* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:49]:

On 2017-09-18 15:39, Patrick Shanahan wrote:

...

* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:24]:

...

Hi,

I need more ram than my 8GiB, but my board doesn't allow it. I use swap, but Leap is very slow when swapping; I have the feeling that it is slower than other releases.

So I wonder if there is some type of small hard disk that can be used for fast swap. SSD? Would it wear out? SSD is typically slow at writing.

ssd would be much faster for swap than revolving rust.

But often not for write, and has a limited number of ops.

I have a machine with an SSD for main disk and feels slow.

...

...

A RAM based "disk"? No need for battery backup, unless to keep the format. Does it exist?

if you are short on ram, how would you justify allocating ram for swap? yes you can do that.

No. On a card, forming a disk. New hardware.

...

but...

my intel based mb specifys 24gig ram but I have 32 and from two different specs. this boundary spec is not always updated when motherboards/bioses receive changes. you could *try* more ram

No, I remember the specs, max is 8.

you misunderstand. the "specs" are not always correct for your mb, usually the max which matches all, not necessarily the max for you mb. try more, it will complain or fault if not usable. try more ram. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-18 17:06, Knurpht - Gertjan Lettink wrote:

Op maandag 18 september 2017 15:56:59 CEST schreef Patrick Shanahan:

...

...

No, I remember the specs, max is 8.

So was my old server/workstation. A BIOS update and now the max is 16 GB.

Huh. The MB has some years and I don't remember updates available. If there were updates for the amount of RAM, I'd noticed. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))

On Mon, Sep 18, 2017 at 9:56 AM, Patrick Shanahan <paka@opensuse.org> wrote:

* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:49]:

...

On 2017-09-18 15:39, Patrick Shanahan wrote:

...

* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:24]:

...

Hi,

I need more ram than my 8GiB, but my board doesn't allow it. I use swap, but Leap is very slow when swapping; I have the feeling that it is slower than other releases.

So I wonder if there is some type of small hard disk that can be used for fast swap. SSD? Would it wear out? SSD is typically slow at writing.

ssd would be much faster for swap than revolving rust.

But often not for write, and has a limited number of ops.

I have a machine with an SSD for main disk and feels slow.

Is this a desktop? If so, for ~$20 you can buy a NVMe SSD controller card. It fits into a PCIe slot. (Assuming you have an empty slot). NVMe SSDs scream. Mine benchmarks above 1.5 GB/sec. (That's B for Bytes, not b for bits). I have openSUSE 42.2 running on it. But I have plenty of RAM, so I don't think the PC ever swaps. Not as fast as RAM, but 10x the speed of rotating drives even if you ignore seek times. I have 2 of the NVMe SSDs. One was way expensive (2 TB). But the Samsung PM951 128GB is under $100. Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Mon, Sep 18, 2017 at 5:27 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

On 2017-09-18 18:28, Greg Freemyer wrote:

...

Is this a desktop? If so, for ~$20 you can buy a NVMe SSD controller card. It fits into a PCIe slot. (Assuming you have an empty slot).

I have free slots, yes, but I don't remember the kind.

...

NVMe SSDs scream. Mine benchmarks above 1.5 GB/sec. (That's B for Bytes, not b for bits). I have openSUSE 42.2 running on it. But I have plenty of RAM, so I don't think the PC ever swaps.

Not as fast as RAM, but 10x the speed of rotating drives even if you ignore seek times.

I have 2 of the NVMe SSDs. One was way expensive (2 TB).

But the Samsung PM951 128GB is under $100.

Interesting, very interesting. Does one need both a controller card and the SSD?

Yes, you have to have both a NVMe SSD controller and a NVMe SSD. The last 2 MBs I bought had integrated NVMe SSD controllers, so for them I can just buy the SSD and plug it in directly to the MB. I can also boot off of the NVMe SSD. I don't think you can boot off the card based controllers. Here's a picture so you know what you're looking for: https://www.amazon.com/dp/B01M8IF0FF/ref=asc_df_B01M8IF0FF5175913/?tag=hyprod-20&creative=395033&creativeASIN=B01M8IF0FF&linkCode=df0&hvadid=198093101467&hvpos=1o1&hvnetw=g&hvrand=12899413238259457045&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9010924&hvtargid=pla-350534282544 I don't know the above vendor, but there should be various sellers of a controller card.

The thing is, when I restore from hibernation I see apps swapping in (back to ram), recovering. For instance, Thunderbird was recovering at 600K/S, according to iotop. Total disk flow was under 2MB/S, according to gkrellm. I'm this instant using the laptop, which disk is capable of about 50 MB/S. Why does Leap recover so slowly from hibernation? It takes minutes till I can really use the machine. 13.1 did the same thing way much faster.

I don't know. Swapping is not something I see much. My smallest machine is 16GB I think. Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-19 00:09, Greg Freemyer wrote:

On Mon, Sep 18, 2017 at 6:00 PM, Greg Freemyer <greg.freemyer@gmail.com> wrote:

...

On Mon, Sep 18, 2017 at 5:27 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:

...

On 2017-09-18 18:28, Greg Freemyer wrote:

...

Is this a desktop? If so, for ~$20 you can buy a NVMe SSD controller card. It fits into a PCIe slot. (Assuming you have an empty slot).

I have free slots, yes, but I don't remember the kind.

...

NVMe SSDs scream. Mine benchmarks above 1.5 GB/sec. (That's B for Bytes, not b for bits). I have openSUSE 42.2 running on it. But I have plenty of RAM, so I don't think the PC ever swaps.

Not as fast as RAM, but 10x the speed of rotating drives even if you ignore seek times.

I have 2 of the NVMe SSDs. One was way expensive (2 TB).

But the Samsung PM951 128GB is under $100.

Interesting, very interesting. Does one need both a controller card and the SSD?

Yes, you have to have both a NVMe SSD controller and a NVMe SSD.

The last 2 MBs I bought had integrated NVMe SSD controllers, so for them I can just buy the SSD and plug it in directly to the MB. I can also boot off of the NVMe SSD. I don't think you can boot off the card based controllers.

Here's a picture so you know what you're looking for:

https://www.amazon.com/dp/B01M8IF0FF/ref=asc_df_B01M8IF0FF5175913/?tag=hyprod-20&creative=395033&creativeASIN=B01M8IF0FF&linkCode=df0&hvadid=198093101467&hvpos=1o1&hvnetw=g&hvrand=12899413238259457045&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9010924&hvtargid=pla-350534282544

I don't know the above vendor, but there should be various sellers of a controller card.

I forgot to say that you can get both NVMe and AHCI SSDs in the M.2 form factor.

You want to get the NVMe if possible.

As far as I can tell the AHCI version were a short lived stepping stone to the full NVMe version. Others might correct me on the details, but regardless go with NVMe.

I see that the cards are pretty expensive. I see this, though: Samsung 960 EVO NVMe M.2 SSD PCI-e 250GB 134,95€ https://www.pccomponentes.com/samsung-960-evo-nvme-m2-ssd-pci-e-250gb The link is in Spanish, but you can see the photos. I understand this things connects directly to a "PCI Express 3.0" interface. I'll have to check whether I have that. There is one on plain PCI, but very expensive: https://www.pccomponentes.com/kingston-hyperx-predator-m2-ssd-480gb-adaptado... Kingston HyperX Predator M.2 SSD 480GB + Adaptador PCIe 344,59€ It is a combo with a PCI to PCI-e interface, then an SSD in M.2 form factor. Another solution, but expensive. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-21 20:32, Greg Freemyer wrote:

On Thu, Sep 21, 2017 at 11:47 AM, Carlos E. R. <> wrote:

...

...

...

I forgot to say that you can get both NVMe and AHCI SSDs in the M.2 form factor.

You want to get the NVMe if possible.

As far as I can tell the AHCI version were a short lived stepping stone to the full NVMe version. Others might correct me on the details, but regardless go with NVMe.

I see that the cards are pretty expensive.

Define expensive in this context.

My initial look pointed at 300€ for the card alone. Probably I entered a bad search term or something, I can't replicate it now.

I think I have this one (~$20): https://www.asus.com/us/Motherboard-Accessory/HYPER_M2_X4_MINI_CARD/

Ah, interesting. .-)

...

I see this, though:

Samsung 960 EVO NVMe M.2 SSD PCI-e 250GB 134,95€

https://www.pccomponentes.com/samsung-960-evo-nvme-m2-ssd-pci-e-250gb

The link is in Spanish, but you can see the photos. I understand this things connects directly to a "PCI Express 3.0" interface. I'll have to check whether I have that.

My x99 MBs (24-months or newer) have a small connector to plug that into. One has the SSD lay parallel to the MB. One has it stand up vertically away from the MB.

http://www.legitreviews.com/wp-content/uploads/2014/08/m2-slot-645x443.jpg

But those MBs accept 64GB and 128GB of ram respectively. I will be shocked if your MB that only takes 8MB of ram has a native M.2 connector.

Me too, but I still need to dig out the MB specs. Wait, I may have a pdf... Yes. The board is a "MSI P45 Diamond" In the photo, it seems to have it or similar. But... It has: - 2 PCI Express x16 slots compatible with PCIE 2.0 spec a. for CrossFire mode, please install both graphics cards on both PCIE x16 slots b. to use 2 PCIE x16 slots, the PCIE x 16 lanes will auto arrange from x16/ x0 to x8/ x8 - 2 PCI Express x 1 slots - 2 PCI slots ... PCI (Peripheral Component Interconnect) Express Slot The PCI Express slot supports the PCI Express interface expansion card. The PCI Express 2.0x 16 supports up to 8.0 GB/s transfer rate. The PCI Express x1 supports up to 250 MB/s transfer rate. ... "The interface of X-Fi Xtreme Audio Card is PCI-E x1." ie, it is used for audio. And the PCIE 2.0 x16 is used for graphics (it supports ATi CrossFire (Multi-GPU), two cards). I have one nvidia. No, I don't think it has it. What you say next applies:

I think you will need something like: https://www.pccomponentes.com/silverstone-ecm21-adaptador-m2-a-pcie-x4

Very interesting find, thanks!

Look at the picture. There are no active components. Basically all that board does is convert a normal PCIe MB connector to the M.2 connector used by NVMe SSDs.

Yes.

If the adapter says it has "M.2 SATA" support, run and hide. You want M.2 PCIe or M.2 NVMe support.

There is a note about that in the specs. It says (I translate): "M.2 PCIe-NVME needs Intel 9 series (Z97 H97 Z170 X99) or a MB with superior chipset with Windows 8 or superior operating system." I think they have translated it from some other language. They also warn that it may not boot, ask the board manufacturer. Manufacturer link: http://www.silverstonetek.com/product.php?pid=703&area=en

To boot off of a M.2 NVMe SSD requires bios support in the MB. Using it as a data drive doesn't as far as I know. (But I've only tried it with newer MBs.)

Ah... it won't do, then, because I am thinking of using that disk to hold the system.

...

There is one on plain PCI, but very expensive:

https://www.pccomponentes.com/kingston-hyperx-predator-m2-ssd-480gb-adaptado...

Kingston HyperX Predator M.2 SSD 480GB + Adaptador PCIe 344,59€

It is a combo with a PCI to PCI-e interface, then an SSD in M.2 form factor.

Another solution, but expensive.

It's primarily expensive because of "480GB". If you want that much capacity it will cost you well above $100. If you're only doing this to address swap space, you should be able to get it done with a 128GB SSD. Around $100 for the SSD / adapter card pair.

Well, if I go for one, it will hold swap and system, not data. It will improve things more. But I don't need that much space for that. Another possibility is dcache or similar. I will then go for a plain SSD SATA disk. 100..200 GB range. I will then have to consider what port I have free or that I can make free. Thank you! :-) -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 22/09/17 13:22, Carlos E. R. wrote:

Intel SSD M.2 Optane 16GB 52,54€ Intel SSD M.2 Optane 32GB 86,54€ Intel SSD M.2 Optane 32GB 80,54€, reconditioned.

This means that the Intel ones fail often enough. The other two suitable items are:

To my mind, recon just means "removed from a 2nd-hand system for use as spares". Intel in particular, I think if it's failed it's bin fodder - they commit suicide and are totally dead. Not worth salvaging. Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 22/09/17 09:26 AM, Carlos E. R. wrote:

I'm guessing at bad solder or connector issue. Not chip failure.

Historically, ever since electrical circuits were put together by soldering wired together, and that's a *VERY* long time, solder joint failure has statistically exceeded component failure. Automated "flow" soldering was a great improvement over hand soldering of PCBs, but the baseline hasn't changed much. There are just too many ways a solder joint can fail. When I googled for this I was overwhelmed by the number of research papers, the amount of research done on this subject. This is indicative of both its significance and of its occurrence; that rate of occurrence is why it is so significant and such an important subject for study. https://en.wikipedia.org/wiki/Failure_of_electronic_components#Contact_failu... -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 22/09/17 11:50 AM, Stevens wrote:

...

Anton - with the advent of surface mount and the Chinese penchant for playing the "I can do this with less than you" game, sometimes all you have to do is glare at a pc board and the solder connection will break. Their reluctance to use expensive (HAH!) solder isn't limited to boards, either. Sometimes cable failure analysis finds no solder at all. Manufacturers rarely use statistical sampling of output; nowdays they look at the returns percentages and sales numbers. Whatta world, whatta world.

Back in my pre-teens I could take apart old radios and recover components: resistors, potentiometers, capacitors -- static or variable, and vales, and build new stuff with them. Modularity means you can't do that; the "maker" culture that is growing up in places like China has to use complete boards or even complete 'things' - re-purposing whole cell phones or more. It is just a different kind of creativity. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-22 19:47, Larry Stotler wrote:

On Fri, Sep 22, 2017 at 11:50 AM, Stevens <fred-n-sandy@myrhinomail.com> wrote:

...

Anton - with the advent of surface mount and the Chinese penchant for playing the "I can do this with less than you" game, sometimes all you have to do is glare at a pc board and the solder connection will break. Their reluctance to use expensive (HAH!) solder isn't limited to boards, either. Sometimes cable failure analysis finds no solder at all. Manufacturers rarely use statistical sampling of output; nowdays they look at the returns percentages and sales numbers. Whatta world, whatta world.

Lead free solder is a big part of the problem. It's no coincidence that gpu chips started pulling off when they made the switch. HP resorted to using epoxy at one point to try to hold the gpus to the motherboard, which made it almost impossible to re-ball the chip. Made a bunch redoing the gpus years ago with regular solder and rarely had any come back.

Lead free solder means higher soldering temperature, for starters. Worst, controlled temperature. And more expensive if silver is added. Higher temperature means that the old PCBS can't stand the heat. Soldering stations are also different, they need to control a precise temperature. If silver is a component of the solder, then one applies the solder as if one is Ebenezer Scrooge. Lead free solder is not mandatory, but recommended. However, many institutions may contract for "lead free" solder, and then you have to comply. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 22/09/17 14:26, Carlos E. R. wrote:

On 2017-09-22 15:24, Wols Lists wrote:

...

On 22/09/17 13:22, Carlos E. R. wrote:

...

Intel SSD M.2 Optane 16GB 52,54€ Intel SSD M.2 Optane 32GB 86,54€ Intel SSD M.2 Optane 32GB 80,54€, reconditioned.

This means that the Intel ones fail often enough. The other two suitable items are:

To my mind, recon just means "removed from a 2nd-hand system for use as spares".

Intel in particular, I think if it's failed it's bin fodder - they commit suicide and are totally dead. Not worth salvaging.

I'm guessing at bad solder or connector issue. Not chip failure.

Is that really seriously economic??? I see a lot of recon gear at Morgan Computers (their IJT Direct subsidiary gets pretty much all my printer business), and recon appears to mean "we buy lots of redundant, surplus and ex-lease gear, QA it, and sell it on". If a computer fails QA they may well strip it for parts, QA those and sell them on. But seriously, if a part fails QA, is it really worth repairing? Collecting dead parts in the printer industry (ink cartridges etc) makes sense because of the sheer volume. But does it really make sense anywhere else? Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 22/09/17 09:58 AM, Wols Lists wrote:

Is that really seriously economic???

I see a lot of recon gear at Morgan Computers (their IJT Direct subsidiary gets pretty much all my printer business), and recon appears to mean "we buy lots of redundant, surplus and ex-lease gear, QA it, and sell it on". If a computer fails QA they may well strip it for parts, QA those and sell them on.

But seriously, if a part fails QA, is it really worth repairing?

Collecting dead parts in the printer industry (ink cartridges etc) makes sense because of the sheer volume. But does it really make sense anywhere else?

it depends. The Deeming Principles for QA of mass production have changes a lot. Process improvements in production, quality improvements in production, have made QA either very statistical or completely pointless. If, when there are statistics, the failure rate is down in the 10^-6, why bother? Along another axis, there is the issue of the whole delivery process line. If the cost of the component or assembly is low enough compared to the cost of repair, then don't bother with repair, just replace. Often without fault analysis. There is a short film strip I saw once that was meant to illustrate something else but does demonstrate this principle. A fighter jet lands on an aircraft carrier and the ground crew flock around it. One team carried refuelling hoses. Another team comes along in phases. The first phase is a guy with a "key" who runs past opening the bay doors along the side of the aircraft that contain the various electronics and other stuff. Immediately following him is a guy with a trolley tethered to his waist who pulls the units out of the pods and chicks them, regardless of their operating status, onto the trolley. Behind him is another guy, again with a trolley, but this is stacked with racks, all in just the right order, and he stuffs them into the slots just vacated. Behind him is a guy that makes sure the racks are all locked in, closes the pod doors and has a key to lock the doors. Meanwhile other teams have refuelled, reloaded the munitions, and the plane is ready to go. The movie was _meant_ to illustrate a point about 'line replaceable units'. The electronics pods didn't have to be single sourced if they had the same interface and function. What it _did_ show is that it isn't worth stopping to test them. Chuck them on the trolley and later take them down to the shop and test them. How is this relevant here? it is not worth disrupting the production like to test every item. At some point its not worth even doing statistical tests. If the competent fails for the customer replace it without question. It's cheaper that way. So: two things emerged from this for me. One was what I term "The Closet of Anxieties". Many firms using Microsoft found that they were upgrading perfectly capable hardware simply because the Windows upgrade demanded new hardware. They didn't have the heart to add the old machines to landfill. One example: A friend of a friend told me about a law office that went though this 'upgrade'. I visited to see what they had. The office manager showed me a room with the "old" (but still fully functional apart from their wiped disks) machines. It was like walking into a graveyard. They were lined up, perhaps two dozen, with the display, looking like headstones. "How much?" I asked. "Nothing. They're free. They are CCA depreciated and off the books now. They are just taking up space. Take one, take two take them all", she said. I could only fit about 10 in the back of the Volvo. It was the monitors that were the problem. I put Linux on them, threw a garden party "with internet service" (it was still a novelty then) and gave them away. Another example: I upgraded to a tower. It had once been a server in the CoA at a place I worked, but the disk was damaged so I bought a 1T drive for a discount store on College. It failed spectacularly the first day. I took it back and they replaced it without even listening to my report on the number of bad sectors I found. I asked what it this one failed. "Bring it back, we;'ll replace it, no charge". That's the economics of it all. I ran badblocks mercilessly on this one! 'Reconditioned' may fail, but so might 'new'. It's the economic model behind production that makes the replacement policy what it is. Yes they could do better post-production testing. But that would be expensive. And we'd pay the cost, since it would apply to EVERY disk drive. It is cheaper all round to do it this way. And, as a side effect, 'reconditioned' drives are cheap. In once sense they've already been paid for. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Out of curiosity, the "Samsung 960 EVO" is rated at 3200 MB/s (read) and 1500 MB/s (write). The SATA interface can support that transfer speed, but not SSD disks (~500 MB/s). Maybe I'm reading it wrong.

I need an aspirin.

The SATA spec is 6.0 Gb/sec. The 960 EVO is 3.2 GB/sec Gb != GB Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-18 18:28, Greg Freemyer wrote:

On Mon, Sep 18, 2017 at 9:56 AM, Patrick Shanahan <paka@opensuse.org> wrote:

...

* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:49]:

...

On 2017-09-18 15:39, Patrick Shanahan wrote:

...

* Carlos E. R. <robin.listas@telefonica.net> [09-18-17 09:24]:

...

Hi,

I need more ram than my 8GiB, but my board doesn't allow it. I use swap, but Leap is very slow when swapping; I have the feeling that it is slower than other releases.

So I wonder if there is some type of small hard disk that can be used for fast swap. SSD? Would it wear out? SSD is typically slow at writing.

ssd would be much faster for swap than revolving rust.

But often not for write, and has a limited number of ops.

I have a machine with an SSD for main disk and feels slow.

Is this a desktop? If so, for ~$20 you can buy a NVMe SSD controller card. It fits into a PCIe slot. (Assuming you have an empty slot).

NVMe SSDs scream. Mine benchmarks above 1.5 GB/sec. (That's B for Bytes, not b for bits). I have openSUSE 42.2 running on it. But I have plenty of RAM, so I don't think the PC ever swaps.

Not as fast as RAM, but 10x the speed of rotating drives even if you ignore seek times.

I have 2 of the NVMe SSDs. One was way expensive (2 TB).

But the Samsung PM951 128GB is under $100.

I think I will start by "improving" my laptop. It has one 500 GB (not GiB) hard disk, dual boot, so I need one of same size or larger. But not much larger because of price. My almost local shop has these - what do you think, any quick "not that one!"? Something you know I should know? I think I might go for the first one on the list, I can afford that price, I think. They are expensive! Crucial MX300 SSD 525GB 150,95€ **** maybe this one random write 92k / 83k IOPS secuencial write: 510 MB/s Samsung 850 Evo SSD Series 500GB SATA3 154,95€ 90,000 IOPS(500GB) 520 MB/s sec write. WD Blue SSD 500GB SATA3 154,59€ (IOPS) 80000 (MB/s) 525 Samsung 850 Pro SSD Series 512GB 245,95€ (4KB) 36000 IOPS 520 MB/s Crucial MX300 SSD 1050GB 283€ IOPS up to 83k 510 MB/s Kingston SSDNow KC400 512GB SATA3 205,59€ SanDisk Extreme PRO SSD 960GB 399€ *** Not this one. Samsung PM871A SSD 512GB SATA3 OEM 182,59€ OEM? 88 KIOPS 515 MB/s SanDisk Ultra II SSD 960GB SATA3 299,59€ -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On Wednesday, 20 September 2017 11:45:46 ACST Carlos E. R. wrote:

[...] I think I will start by "improving" my laptop. It has one 500 GB (not GiB) hard disk, dual boot, so I need one of same size or larger. But not much larger because of price.

My almost local shop has these - what do you think, any quick "not that one!"? Something you know I should know?

I can vouch for the Samsung EVO 850 SSD’s. I have one in my desktop machine and it hasn’t missed a beat. Also, their sizes are closer to those of traditional rotating rust drives, unlike some other SSD’s. Haven’t had so much luck with Kingston SSD’s (which I find a little surprising given that Kingston RAM used to be the go-to for fussy hardware). The Samsung 850 Pro should also be good, if the Evo series is anything to go by. I don’t know enough about the Crucial, San Disk or WD offerings to comment. Regards, Rodney. -- ============================================================== Rodney Baker VK5ZTV rodney.baker@iinet.net.au CCNA #CSCO12880208 ============================================================== -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 09/18/2017 02:20 PM, Carlos E. R. wrote:

On 2017-09-18 15:56, Patrick Shanahan wrote:

...

* Carlos E. R. <> [09-18-17 09:49]:

...

...

...

but...

my intel based mb specifys 24gig ram but I have 32 and from two different specs. this boundary spec is not always updated when motherboards/bioses receive changes. you could *try* more ram

No, I remember the specs, max is 8.

you misunderstand. the "specs" are not always correct for your mb, usually the max which matches all, not necessarily the max for you mb. try more, it will complain or fault if not usable. try more ram.

The specs of my mb specify 8 MiB. I'm not going to spend the money on more with a 95% chance of wasting it.

If I had it on a drawer, I'd try.

There is a company called Crucial.com in the US that sells memory. Long ago I called them asking how much upgrade to the ram i could get a machine. They started out saying Officially you can get XXX. I asked what do you mean Officially. They said that is all the motherboard manufacture said it would take. But we will sell you XXX+yyy and we will guarentee it will work or your money back for the yyy part. It worked. They won't always do this. Some mother boards will and some won't take more. -- After all is said and done, more is said than done.

Op dinsdag 19 september 2017 04:09:35 CEST schreef Doug:

On 09/18/2017 07:19 PM, John Andersen wrote:

...

On 09/18/2017 02:20 PM, Carlos E. R. wrote:

...

On 2017-09-18 15:56, Patrick Shanahan wrote:

...

* Carlos E. R. <> [09-18-17 09:49]:

...

...

but...

my intel based mb specifys 24gig ram but I have 32 and from two different specs. this boundary spec is not always updated when motherboards/bioses receive changes. you could *try* more ram

No, I remember the specs, max is 8.

you misunderstand. the "specs" are not always correct for your mb, usually the max which matches all, not necessarily the max for you mb. try more, it will complain or fault if not usable. try more ram.

The specs of my mb specify 8 MiB. I'm not going to spend the money on more with a 95% chance of wasting it.

If I had it on a drawer, I'd try.

There is a company called Crucial.com in the US that sells memory. Long ago I called them asking how much upgrade to the ram i could get a machine. They started out saying Officially you can get XXX. I asked what do you mean Officially. They said that is all the motherboard manufacture said it would take. But we will sell you XXX+yyy and we will guarentee it will work or your money back for the yyy part. It worked.

They won't always do this. Some mother boards will and some won't take more. This is correct. I have an old Dell laptop--Inspiron E1505--that Dell said only can use 2 GiB. But you can put 4 gig in it, and when you do, it will only use 3.2. That's an improvement, of course, but apparently the mobo is hard-wired so as not to be able to use the entire 4. --doug A 32bit machine? That would explain the use of only 3.2 GB

-- Gertjan Lettink, a.k.a. Knurpht openSUSE Board Member openSUSE Forums Team -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, Sep 19, 2017 at 9:56 AM, Bengt Gördén <bengan@bag.org> wrote:

Den 2017-09-19 kl. 06:25, skrev Knurpht - Gertjan Lettink:

...

A 32bit machine? That would explain the use of only 3.2 GB

Never heard that one before.

https://en.wikipedia.org/wiki/PCI_hole -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Bengt Gördén wrote:

Den 2017-09-19 kl. 09:10, skrev Andrei Borzenkov:

...

On Tue, Sep 19, 2017 at 9:56 AM, Bengt Gördén <bengan@bag.org> wrote:

...

Den 2017-09-19 kl. 06:25, skrev Knurpht - Gertjan Lettink:

...

A 32bit machine? That would explain the use of only 3.2 GB

Never heard that one before.

https://en.wikipedia.org/wiki/PCI_hole

Now that you mention it. My bad. I've actually both heard of it and experienced it. Long time ago. We built the first root-dns (i.root) on x86-hardware many years ago (around 1998-1999). My college at that time made some remarkably linguistic acrobatic when he found out that this was a limit.

At that time, PAE was the solution. (as of Pentium Pro I believe). -- Per Jessen, Zürich (11.2°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 09/19/2017 01:56 AM, Bengt Gördén wrote:

Den 2017-09-19 kl. 06:25, skrev Knurpht - Gertjan Lettink:

...

A 32bit machine? That would explain the use of only 3.2 GB

Never heard that one before.

2^32=4294967296

Anyway, it now has a 64-bit processor and an upgraded BIOS to go with it, but it still only accesses 3.2 GiB. --doug -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 09/19/2017 11:35 AM, Andrei Borzenkov wrote:

19.09.2017 21:18, Doug пишет:

...

On 09/19/2017 01:56 AM, Bengt Gördén wrote:

...

Den 2017-09-19 kl. 06:25, skrev Knurpht - Gertjan Lettink:

...

A 32bit machine? That would explain the use of only 3.2 GB

Never heard that one before.

2^32=4294967296

Anyway, it now has a 64-bit processor and an upgraded BIOS to go with it, but it still only accesses 3.2 GiB.

Does it also have 64 bit OS?

Yes, I'm sure it does. I have a similar machine (inspiron 9400) its another older laptop that still performs well. I have the same 3.2 gig usable in that machine, even thought 4 gig are installed.

jsa@Inspiron1:~> uname -ar Linux Inspiron1 4.4.87-18.29-default #1 SMP Wed Sep 13 07:07:43 UTC 2017 (3e35b20) x86_64 x86_64 x86_64 GNU/Linux

Dell Core 2 Duo machine. Nobody is exactly sure how Dell did this (or why) but these machines and the 1501 series all have this limitation. Some say its a bios limitation that was removed with bios 2.6.3. See http://en.community.dell.com/support-forums/laptop/f/3518/p/19298222/1963901... which is specific to the above mentioned Dell 1501 machine. I'm moving that machine to LXDE soon, but maybe I'll try the bios upgrade first. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, Sep 19, 2017 at 2:35 PM, Andrei Borzenkov <arvidjaar@gmail.com> wrote:

...

Anyway, it now has a 64-bit processor and an upgraded BIOS to go with it, but it still only accesses 3.2 GiB.

Does it also have 64 bit OS?

Doesn't matter. My Thinkpad T60p has an Intel 945 series chipset. It was originally released for the 32bit Core Duo(Socket M), but it's been upgraded to a T7600G Core2 Duo. I have installed 4GB (2x2GB) but it only sees 3GB. I think that's a BIOS limitation by Lenovo. So I now run 3GB. My son has a Dell XPS 1710 with a T7600 and has 4GB. But it can only see 3.25GB RAM. That's because it has a 512MB nVidia GeForce GO 7900GTX. It also has a 945 series chipset(Socket M). There were a few motherboards that could map the video RAM and other stuff above 4GB, but it caused issues with Windows, so it wasn't popular. See here: https://en.wikipedia.org/wiki/PCI_hole#Mapping_memory_to_addresses_above_4_G... -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 18/09/17 15:33, Lew Wolfgang wrote:

...

So I wonder if there is some type of small hard disk that can be used for fast swap. SSD? Would it wear out? SSD is typically slow at writing.

We've been using SSD's for root and swap partitions on scientific workstations and servers since 2008 or so, and we've never had a problem. Indeed, the first one I installed is still working after running 24/7 for nine years. Writing to SSD's may be slower than reading, but the writing is still MUCH faster than writing to spinning disks. I'd give it a try, Carlos. Your motherboard does support SATA, right?

It's been mentioned before - the "test to destruction" test on SSDs. I think it took about 18 months of a machine locked into a "write, read, rewrite, erase, reformat, restart" loop before the drives even started thinking of failing. If your PC only has 8GB of ram, are you really going to thrash the drive that hard? The other thing is, depending on the PC, might it be simpler to just upgrade the mobo? I guess you're on DDR2? My DDR3 mobo is 16GB. And I'm looking at upgrading - the new setup will take 4 x 8 or 16 GB sticks, and the cost will be about £150 plus ram (£50 for the mobo, £100 for the processor, ram maybe £30 a stick?) My current machine is probably a good five years old, if not more ... Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 09/18/2017 08:28 AM, Wols Lists wrote:

It's been mentioned before - the "test to destruction" test on SSDs. http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-d...

They all lasted orders of magnitude beyond their warranted/designed life time. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-18 23:43, Carlos E. R. wrote:

On 2017-09-18 17:28, Wols Lists wrote:

...

The other thing is, depending on the PC, might it be simpler to just upgrade the mobo? I guess you're on DDR2? My DDR3 mobo is 16GB.

I don't remember, and "hwinfo --memory" does not say. I think it is DDR3.

Quad-Kit DIMM 8 GB DDR3-1333, looking at the purchase order (Oct 2009). -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))

On 2017-09-19 14:51, David C. Rankin wrote:

On 09/18/2017 04:59 PM, Carlos E. R. wrote:

...

On 2017-09-18 23:43, Carlos E. R. wrote:

...

On 2017-09-18 17:28, Wols Lists wrote:

...

The other thing is, depending on the PC, might it be simpler to just upgrade the mobo? I guess you're on DDR2? My DDR3 mobo is 16GB.

I don't remember, and "hwinfo --memory" does not say. I think it is DDR3.

Quad-Kit DIMM 8 GB DDR3-1333, looking at the purchase order (Oct 2009).

# dmidecode will show info about each slot and the ram controller. It gives a pretty clear shapshot of the hardware (from the hardware).

Stepping back a bit -- why do you need more RAM? Are you trying to keep a copy of all satellite maps in RAM at once? I get that large database apps and other server apps will use as much as you have, but with a base OS running Win7 virtualized and Arch virtualized on the same server, I barely crack 5GiB and that's allocating 2G to each guest OS?

Because my system is swapping actively. As I go from workspace to workspace I see the application that I have there (say, firefox, libreoffice) wake up, and sometimes this takes minutes.

ssd will help. Is there any way to optimize what you have running to play better in the RAM you have?

Unload apps, of course. Firefox is huge. So is Thunderbird. Then I also use LibreOffice, also large. Interestingly, "clamd" is a very large app in RAM, and it does not swap out. Half a gigabyte. I have pending to work out how to load it on demand, and unload manually or automatically when not posting. This moment I'm using vmware player, takes 2 gigs (for Windows 10). -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

* Dave Howorth <dave@howorth.org.uk> [09-19-17 09:36]:

On Tue, 19 Sep 2017 15:18:28 +0200 "Carlos E. R." <robin.listas@telefonica.net> wrote:

...

On 2017-09-19 14:51, David C. Rankin wrote:

...

Stepping back a bit -- why do you need more RAM? Are you trying to keep a copy of all satellite maps in RAM at once? I get that large database apps and other server apps will use as much as you have, but with a base OS running Win7 virtualized and Arch virtualized on the same server, I barely crack 5GiB and that's allocating 2G to each guest OS?

Because my system is swapping actively. As I go from workspace to workspace I see the application that I have there (say, firefox, libreoffice) wake up, and sometimes this takes minutes.

...

ssd will help. Is there any way to optimize what you have running to play better in the RAM you have?

Unload apps, of course.

Firefox is huge.

< 500 MB on my machine

...

So is Thunderbird.

Claws is a lot smaller

...

Then I also use LibreOffice, also large.

< 250 MB

My system is currently using < 4 GB

<from systemmonitor> Process 28513 - firefox Summary The process firefox (with pid 28513) is using approximately 3.8 GB of memory. It is using 3.7 GB privately, 3.3 MB for pixmaps, and a further 295.2 MB that is, or could be, shared with other programs. Dividing up the shared memory between all the processes sharing that memory we get a reduced shared memory usage of 138.6 MB. Adding that to the private and pixmap usage, we get the above mentioned total memory footprint of 3.8 GB. Library Usage The memory usage of a process is found by adding up the memory usage of each of its libraries, plus the process's own heap, stack and any other mappings, plus the stack of its 83 threads. Private more 3741532 KB [heap] 37960 KB /usr/lib64/firefox/libxul.so 25728 KB /SYSV00000000 (deleted) 16740 KB /dev/nvidiactl 2356 KB /usr/lib64/libnvidia-glcore.so.384.69 Shared more 113396 KB /SYSV00000000 (deleted) 8104 KB /dev/shm/org.chromium.m3g8Ek (deleted) 8104 KB /dev/shm/org.chromium.8s9IqF (deleted) 8104 KB /dev/shm/org.chromium.n84qvd (deleted) 8104 KB /dev/shm/org.chromium.ZiLV7t (deleted) Totals Pixmap 3330 KB (Might be stored in the graphics card's memory) Private 3836616 KB (= 227860 KB clean + 3608756 KB dirty) Shared 302288 KB (= 30348 KB clean + 271940 KB dirty) Rss 4138904 KB (= Private + Shared) Pss 3978528 KB (= Private + Shared/Number of Processes) Swap 0 KB </from systemmonitor> but I have 32gig resident and no swap. and I know Carlos keeps several firefox sessions open. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 20 Sep 2017 03:53:01 +0200 "Carlos E. R." <robin.listas@telefonica.net> wrote:

On 2017-09-19 15:36, Dave Howorth wrote:

...

On Tue, 19 Sep 2017 15:18:28 +0200 "Carlos E. R." <robin.listas@telefonica.net> wrote:

...

On 2017-09-19 14:51, David C. Rankin wrote:

...

Stepping back a bit -- why do you need more RAM? Are you trying to keep a copy of all satellite maps in RAM at once? I get that large database apps and other server apps will use as much as you have, but with a base OS running Win7 virtualized and Arch virtualized on the same server, I barely crack 5GiB and that's allocating 2G to each guest OS?

Because my system is swapping actively. As I go from workspace to workspace I see the application that I have there (say, firefox, libreoffice) wake up, and sometimes this takes minutes.

...

ssd will help. Is there any way to optimize what you have running to play better in the RAM you have?

Unload apps, of course.

Firefox is huge.

< 500 MB on my machine

1,601g in mine (RES on "top"). And I see it growing while I do this, and do nothing in FF. 1,685g now.

I don't know what you've got open but I'd guess you have javascript enabled, and perhaps no ad-blocker.

194156 in swap. 4388012 virt.

...

...

So is Thunderbird.

Claws is a lot smaller

Maybe, but I'm used to it and its features.

I used to use both Thunderbird and Evolution. Eventually they both annoyed me enough to bin them. Claws is similar enough and does what I want, although I'd be the first to admit that it's a little idiosyncratic.

...

...

Then I also use LibreOffice, also large.

< 250 MB

239300 RES currently, with only one opened document. If I open my customary calc sheets, it goes to 410176 instantly, before doing anything on them.

So probably similar to mine. I don't find it unexpected that spreadsheets eat memory. Is gnumeric or other any better?

...

My system is currently using < 4 GB

More or less like mine, considering that 4 Gigs are in cache and buffers and free.

cer@Telcontar:~> free -h --si total used free shared buffers cached Mem: 8,2G 7,7G 458M 185M 287M 3,4G -/+ buffers/cache: 4,0G 4,2G Swap: 43G 1,4G 42G

# free -h --si total used free shared buffers cached Mem: 8.0G 4.6G 3.4G 84M 287M 3.3G -/+ buffers/cache: 1.0G 6.9G Swap: 29G 1.2M 29G -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 20 Sep 2017 12:54:43 +0200 "Carlos E. R." <robin.listas@telefonica.net> wrote:

On 2017-09-20 12:30, Dave Howorth wrote:

...

On Wed, 20 Sep 2017 03:53:01 +0200 "Carlos E. R." <> wrote:

...

...

...

...

Firefox is huge.

< 500 MB on my machine

1,601g in mine (RES on "top"). And I see it growing while I do this, and do nothing in FF. 1,685g now.

I don't know what you've got open but I'd guess you have javascript enabled, and perhaps no ad-blocker.

javascript is of course enabled and

There's no of course at all. I always run with it disabled except for selected sites. Apart from the security aspect JS is one of the prime CPU and memory hogs.

I use AddBlock Plus (Easy List, adblock removal list, malware domains, fanboy's social block list, + Allow some non-intrusive advertising).

I was looking at some amazon pages, and some of my local computer shop. Very commercial sites.

...

...

194156 in swap. 4388012 virt.

...

...

So is Thunderbird.

Claws is a lot smaller

Maybe, but I'm used to it and its features.

I used to use both Thunderbird and Evolution. Eventually they both annoyed me enough to bin them. Claws is similar enough and does what I want, although I'd be the first to admit that it's a little idiosyncratic.

:-)

I also use Alpine.

How well does claws handle html? Both reading and writing.

I don't know. I never write 'HTML mail' I only write mail. I generally only read mail as well although it can display 'HTML mail' well enough on the odd occasion I do want to see it.

...

...

...

...

Then I also use LibreOffice, also large.

< 250 MB

239300 RES currently, with only one opened document. If I open my customary calc sheets, it goes to 410176 instantly, before doing anything on them.

So probably similar to mine. I don't find it unexpected that spreadsheets eat memory. Is gnumeric or other any better?

I don't remember, but it is not up to the task.

Claws corrupted the format of the "free" output below so that it is unreadable.

I think both it and TB wrapped the lines to meet the preferred standard. ISTR there's some way in Claws to tell it not to wrap and follow other weird formatting parts of the standards.

Claws:

...

cer@Telcontar:~> free -h --si total used free shared buffers cached Mem: 8,2G 7,7G 458M 185M 287M 3,4G -/+ buffers/cache: 4,0G 4,2G Swap: 43G 1,4G 42G

Reformated by hand:

...

...

...

My system is currently using < 4 GB

More or less like mine, considering that 4 Gigs are in cache and buffers and free.

cer@Telcontar:~> free -h --si total used free shared buffers cached Mem: 8,2G 7,7G 458M 185M 287M 3,4G>> -/+ buffers/cache: 4,0G 4,2G Swap: 43G 1,4G 42G

Thunderbird also broke this paragraph of yours:

...

# free -h --si total used free shared buffers cached Mem: 8.0G 4.6G 3.4G 84M 287M 3.3G -/+ buffers/cache: 1.0G 6.9G Swap: 29G 1.2M 29G

It should be:

...

# free -h --si total used free shared buffers cached Mem: 8.0G 4.6G 3.4G 84M 287M 3.3G -/+ buffers/cache: 1.0G 6.9G Swap: 29G 1.2M 29G

According to that, I have half a gig free, and you have 3.4. And I have 1.4 G in swap, and you nothing.

And my system runs nicely :) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 09/19/2017 08:18 AM, Carlos E. R. wrote:

Because my system is swapping actively. As I go from workspace to workspace I see the application that I have there (say, firefox, libreoffice) wake up, and sometimes this takes minutes.

...

ssd will help. Is there any way to optimize what you have running to play better in the RAM you have? Unload apps, of course.

Firefox is huge. So is Thunderbird. Then I also use LibreOffice, also large.

Interestingly, "clamd" is a very large app in RAM, and it does not swap out. Half a gigabyte. I have pending to work out how to load it on demand, and unload manually or automatically when not posting.

This moment I'm using vmware player, takes 2 gigs (for Windows 10).

Well, you must be doing something crazy with the number of apps you have open, because I have 12 apps open (15 tabs in FF), Tbird with 1000's of message in 1/2 dozen accounts, 3 sessions of Kate with over 100 files open, 10 tabs in konsole, 2 instances of kwrite, 1 gtkwrite, 2 kpdf, basket notepads, keepassx and I don't even crack 4G of active RAM, $ free -tm total used free shared buffers cached Mem: 7932 5618 2314 28 454 3230 -/+ buffers/cache: 1932 6000 Swap: 2071 0 2071 Total: 10004 5618 4386 ## RAM used less buffers/cache $ echo $((5618-1932)) 3686 $ grep 'Active:' /proc/meminfo Active: 3961272 kB I also have Win7 installed as a virtualbox guest, locally and virtualized on my server. I don't get slow desktop swaps regardless. Though I do prefer running win7 on the server started --headless and accessed via rdesktop (which over wireless lan is fine from laptop is fine). I run libre writer and calc without issue. It seems like your desktop is doing something screwy on a per-desktop cache basis that is causing unneeded paging to swap of the inactive desktop apps resulting in very ssslllloooowww desktop switching. I am bit reluctant to ask which desktop? -- David C. Rankin, J.D.,P.E.

On 2017-09-20 20:03, David C. Rankin wrote:

On 09/19/2017 08:18 AM, Carlos E. R. wrote:

It seems like your desktop is doing something screwy on a per-desktop cache basis that is causing unneeded paging to swap of the inactive desktop apps resulting in very ssslllloooowww desktop switching.

I am bit reluctant to ask which desktop?

XFCE. Actually, I suspect how Leap handles swap. It reads it very slowly. 13.1 was faster, same load. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-18 16:33, Lew Wolfgang wrote:

On 09/18/2017 06:23 AM, Carlos E. R. wrote:

...

Hi,

I need more ram than my 8GiB, but my board doesn't allow it. I use swap, but Leap is very slow when swapping; I have the feeling that it is slower than other releases.

So I wonder if there is some type of small hard disk that can be used for fast swap. SSD? Would it wear out? SSD is typically slow at writing.

We've been using SSD's for root and swap partitions on scientific workstations and servers since 2008 or so, and we've never had a problem. Indeed, the first one I installed is still working after running 24/7 for nine years. Writing to SSD's may be slower than reading, but the writing is still MUCH faster than writing to spinning disks.  I'd give it a try, Carlos.  Your motherboard does support SATA, right?

It does, yes. I might have a think at using one SSD disk as cache for the rotating disks. I have four, the smaller is 2TB. Dunno, it is not easy.

BTW, I installed my latest server a few weeks ago that has 512-GB of ECC RAM, and my users have still managed to fill it up and overflow to swap. Yes, SSD drives for root and swap too.  Leap 42.3 really screams on it!

You got me drooling. ;-) -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))

On 2017-09-18 19:56, John Andersen wrote:

On 09/18/2017 06:23 AM, Carlos E. R. wrote:

Another few points to consider Carlos:

Sometimes you think you are using swap, but you check and find out that the machine is set up by default to resist swapping until it gets desperate for memory. This has been the trend lately but it is a silly default when you have swap on an SSD.

I'm currently running 42.2 on a laptop limited to 4gig. I crank up the swappiness (60) and crank down the cache pressure (50). I WANT it to swap. I want it to swap early and often.

Its using 4 gig of memory with a 6gig swap on SSD in a dual core machine. Sometimes running VMs I'm using more swap space than I am real memory. Its fast as hell.

In /etc/sysctl.conf vm.swappiness = 60 vm.vfs_cache_pressure = 50

I make much more use of swap this way than with the default settings.

Oh, I know I use swap. On the laptop: top - 23:45:17 up 9 days, 10:29, 3 users, load average: 0.19, 0.33, 0.79 Tasks: 285 total, 2 running, 282 sleeping, 0 stopped, 1 zombie %Cpu(s): 6.1 us, 2.1 sy, 0.1 ni, 91.0 id, 0.4 wa, 0.0 hi, 0.1 si, 0.0 st KiB Mem: 3947076 total, 3362332 used, 584744 free, 118484 buffers KiB Swap: 6289412 total, 1458472 used, 4830940 free. 1810508 cached Mem ******** Desktop (rebooted today, so not many tasks) top - 23:46:05 up 8:18, 3 users, load average: 0.59, 0.48, 0.54 Tasks: 462 total, 2 running, 459 sleeping, 0 stopped, 1 zombie %Cpu(s): 11.2 us, 1.8 sy, 0.0 ni, 87.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 8174404 total, 7330220 used, 844184 free, 488952 buffers KiB Swap: 43744244 total, 1211500 used, 42532744 free. 3275364 cached Mem ******* I have seen swap at 4 Gigs. When you hibernate, everything goes to swap. The problem is, Leap recovers very slowly. Takes minutes. changing workplace in XFCE to use LOffice, and I have to wait minutes for it to respond.

Now about your worry regarding write to SSD.... Yes write is slower than read. But only because Read is so blindingly fast. Writing on ssd is light years ahead of HDD write speeds. You can only call it slow when viewed in comparison to the same SSD's read speed.

So don't be afraid to write on your SSD, and don't be afraid to use the hell out of your SSD. You will send the machine to recycle long before your ssd wears out.

Another possible source of slowness: discard option in fstab. You want to remove that from every partition in fstab that sits on ssd. Instead use systemd's fstrim.timer.

Ah.

sudo systemctl enable fstrim.timer sudo systemctl start fstrim.timer sudo systemctl status fstrim.service

This will trim you drives once a week (which is plenty) rather than inserting trim operations after ever file delete operation. Discard in the fstab entry can make the machine "feel" slow and causes excess SSD access.

Interesting.

One recommendation for swap on ssd (or any partition on ssd) is to over provision, provide more space than you actually need so as not to force trim operations for space recovery. Ned 8 gig? then partition for 10.

On the one computer with an SSD, I left space not partitioned.

Some links I've found on this subject:

http://www.linux-magazine.com/Issues/2016/187/SSD-tuning

Somewhat dated: Mostly from Ted Ts'o who knows a thing or two about file systems: https://thunk.org/tytso/blog/2009/02/22/should-filesystems-be-optimized-for-... Newer post by Ted Ts'o on this: https://forums.freebsd.org/threads/56951/#post-328912

Also Less specific: https://arstechnica.com/gadgets/2015/04/ask-ars-my-ssd-does-garbage-collecti...

ArchWiki also recommends against discard: https://wiki.archlinux.org/index.php/Solid_State_Drives

Also, (side issue nothing to do with swap) btrfs is not really optimized for ssd. https://btrfs.wiki.kernel.org/index.php/FAQ#Is_Btrfs_optimized_for_SSD.3F

Thanks for the links. No, I stay away from btrfs. I use ext4, xfs, and some reiserfs. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))

On 09/18/2017 10:18 PM, Anton Aylward wrote:

On 18/09/17 05:51 PM, Carlos E. R. wrote:

...

No, I stay away from btrfs. I use ext4, xfs, and some reiserfs. I've written a number of times about the idiocy of ext4, a B-Tree allocator that has mkfs-time provisioning of a hard boundary between space reserved for files and space reserved for inodes. It's sheer idiocy! This dates back to the very early UNIX, what we later called the V7FS.

It really is inexcusable, since before ext4 came into being we already had not only ReiserFS but also XFS as examples of how to get around the pre-provisioning constraint, to make ALL allocations, data and inodes, dynamically, from the B-Tree.

Then we have BtrFS, which is also a B-Tree file system and also allocates space and inodes dynamically from the B-tree rather than preallocating.

Now it all goes to rat-shit.

openSuse 15 will no longer support ReiserFS. This isn't a passive move, as in "no continued support, no further fixes", but is an active move. If you have any ReiserFS partitions then they will be forcibly converted to another file system whether you want that to happen or not.

This seems very draconian to me.

Yes, I realise that Oracle are going full steam ahead with BtrFS as their anointed FS for their Linux. But what about other distributions?

https://news.ycombinator.com/item?id=14909843 https://forums.theregister.co.uk/forum/1/2017/08/16/red_hat_banishes_btrfs_f...

And an older "10 reasons not to use ZFS" https://www.redhat.com/archives/rhl-list/2006-June/msg03623.html

While BtrFS _could_ have been a replacement for ReiserFS and a 'nicer' version of XFS, the developers went overboard and decided to do a "One File System to rule all storage" approach. Integrate all of what LVM does but without actually partitioning. Have a many other features that annoy people who don't understand it and are only relevant for server class machines and irrelevant to embedded systems or many workstations.

Resiser4 is available for kernel 4.13, which is the latest I have from repository kernel_stable. https://www.phoronix.com/scan.php?page=news_item&px=Reiser4-Linux-4.13 Why can't openSuse work that in, after all, there is a developer/group working on it. Could there be one for openSuse?

I'm not sure what you are suggesting as a universal file system. You are of course aware that Reiser is in prison, probably for the rest of his life, and not supporting the FS he devised, and I haven't heard that anyone else is, unless you are. From your post, I do not see a recommendation for a "file system that fits all." (It would be nice if it worked not only for Linux but for FreeBSD. so that exchanging files between systems would be easy, and hardware --hard drives-- would work together.) --dm -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-19 06:20, Doug wrote:

On 09/18/2017 10:18 PM, Anton Aylward wrote:

...

...

While BtrFS _could_ have been a replacement for ReiserFS and a 'nicer' version of XFS, the developers went overboard and decided to do a "One File System to rule all storage" approach.  Integrate all of what LVM does but without actually partitioning.  Have a many other features that annoy people who don't understand it and are only relevant for server class machines and irrelevant to embedded systems or many workstations.

Resiser4 is available for kernel 4.13, which is the latest I have from repository kernel_stable. https://www.phoronix.com/scan.php?page=news_item&px=Reiser4-Linux-4.13 Why can't openSuse work that in, after all, there is a developer/group working on it. Could there be one for openSuse?

I'm not sure what you are suggesting as a universal file system.

He refers to the move to do it all with btrfs on openSUSE. Internal raid, internal encryption, features similar to LVM also internal.... Anton is not suggesting that move, on the contrary. Others are doing that move, in fact. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

Anton Aylward wrote:

The decision on the part of the openSUSE packagers to not include Reiser4 is understandable; the decision to force us not to be able to use ReiserFS, to force conversion to another FS rather than continue to include it as one of the many options for a kernel module, is outrageous.

I don't know if you are over-interpreting the below, or if you have your information from elsewhere: https://lizards.opensuse.org/2017/08/24/yast-sprint-41/ "The support of new installations with ReiserFS was removed from YaST in SUSE Linux Enterprise 12 and openSUSE Leap 42 but upgrades were still supported." "... the installer will block the upgrade of systems formatted with ReiserFS." The latter I find somewhat odd, but it appears to be YaST-only, not zypper. "If some of the entries in the /etc/fstab file of the system to be upgraded is using ReiserFS, the installer will suggest to convert them to another filesystem type before migrating the system to SUSE Linux Enterprise 15 or openSUSE Leap 15." "A similar blocking error will be reported for ReiserFS root partitions." This is also all YaST-only, and I still find it pretty odd, but it won't prevent anyone continuing to use reiserfs. -- Per Jessen, Zürich (12.5°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-19 13:57, Anton Aylward wrote:

On 19/09/17 07:28 AM, Per Jessen wrote:

...

This is also all YaST-only, and I still find it pretty odd, but it won't prevent anyone continuing to use reiserfs.

it WILL prevent anyone using the installer to upgrade to leap15 if they want to continue using ReiserFS. it looks like there will be no provision in the Leap15 kernel for ReiserFS as a module any more.

Only if it is used on the "/" partition.

ReiserFS never was part of the kernel, it was always a module. If you wanted it in the kernel you had to recompile it yourself. Ditto Reiser4.

But the announcement doesn't say anything about removing the module from the distribution kernel, only from YaST. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-20 00:29, Anton Aylward wrote:

On 19/09/17 08:14 AM, Carlos E. R. wrote:

...

On 2017-09-19 13:57, Anton Aylward wrote:

...

On 19/09/17 07:28 AM, Per Jessen wrote:

...

This is also all YaST-only, and I still find it pretty odd, but it won't prevent anyone continuing to use reiserfs.

it WILL prevent anyone using the installer to upgrade to leap15 if they want to continue using ReiserFS. it looks like there will be no provision in the Leap15 kernel for ReiserFS as a module any more.

Only if it is used on the "/" partition.

No, ANY AND ALL. it says quite clearly that it scans all entries in the /etc/fstab

And then *suggests* to change it. Does not block the upgrade.

...

...

ReiserFS never was part of the kernel, it was always a module. If you wanted it in the kernel you had to recompile it yourself. Ditto Reiser4.

But the announcement doesn't say anything about removing the module from the distribution kernel, only from YaST.

YaST is just a UI front end. You can't remove Reiser fro Yast since it was never in yast in the first place.

Yes, it was. We could, in the past, create reiserfs partitions. The YaST partitioner supported it. Leap does not.

ReiserFS is, at lest up to the kernel I'm running, 4.13.2-1.g68f4aee-default, under 42.2, a module. It never was compiled into a distribution, though heavens knows what the Build Service may have produced! I've always used it as a module.

That reiserfs was and is compiled as a module doesn't mean it is not part of the kernel or that it is built separately. It does not matter, it is an integral part of the kernel. As always, except when it was born. The only difference is that if you do not have any reiserfs partition, you save some kernel memory space.

Now the issue is whether or not Leap-15 removes it as a module. Forget this obsession with yast.

The announcement is only about YaST. Forget what RB said, there is no announcement to remove it from the kernel. I don't think he has that power, as it comes from upstream. At worst, he could disable ('n') its compilation, and then we would have to build our own kernels. A nuisance, certainly. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

Anton Aylward wrote:

On 19/09/17 07:28 AM, Per Jessen wrote:

...

This is also all YaST-only, and I still find it pretty odd, but it won't prevent anyone continuing to use reiserfs.

it WILL prevent anyone using the installer to upgrade to leap15 if they want to continue using ReiserFS.

Judging by that article, that is correct. Fortunately there is still "zypper dup".

it looks like there will be no provision in the Leap15 kernel for ReiserFS as a module any more.

I don't see that mentioned anywhere. It's also difficult to see why we should go to the length of changing the kernel build config to exclude a module just because we don't want to support that file system.

ReiserFS never was part of the kernel, it was always a module.

Uh, gazillions of things are "part of the kernel", as modules.

If you wanted it in the kernel you had to recompile it yourself.

If "in the kernel" means "not as a module", you are right. Same applies to most of the other file systems and drivers. -- Per Jessen, Zürich (13.0°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 19/09/17 09:40 AM, Wols Lists wrote:

On 19/09/17 14:17, Per Jessen wrote:

...

...

ReiserFS never was part of the kernel, it was always a module.

...

Uh, gazillions of things are "part of the kernel", as modules.

Uh, actually, as a *supported* filesystem, Reiser should never have been a module. It's far too easy to hose your system by mistake.

Well DUH! There are a gazillion ways to hose your system using other modules! BTDT. But NEVER EVER with ReiserFS. With BtrFS, yes; with XFS, yes; with ext3, yes. NEVER EVER with ReiserFS. Now I'm sure there are others who will post here that they've had/had-not and problem with {XFS,ext2,ext3,ext4,Reiser,NullFS,overlay,....} What does it prove? one thing and one thing only: That people are fallible Big Deal! We knew that. It is, however, not justification for pulling an established and mature file system. If it is then we should pull sysv, XFS and few others as well.

(running gentoo, I have ext and raid4/5/6 compiled in to my kernel, *not* modules. That means my kernel doesn't need external help to read the root filesystem ...)

And what constitutes "extra help"? There is a "yes, it used to be, but we changed all that". There were a number of file systems that you couldn't boot without "extra help", but that went by the wayside with grub2. Please research a little before making assertions about what others cannot do. I get very upset when people try telling me that I can't do, for a pile of theoretical or outdated reasons, that I can't do something I have been doing.

If you are using Reiser for non-root partitions, you should be able to upgrade and then just recompile Reiser support.

It looks as if they are going to pull from the modules. That's how I'm using it now. I don't need it compiled into the kernel. I want my INITRD to be as small as possible. So it does what it has to as fast a possible to get the core system loaded, Defer as much as possible. This has been discussed endlessly in other threads, and I'm not going into it all again here. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-20 00:34, Anton Aylward wrote:

On 19/09/17 09:17 AM, Per Jessen wrote:

...

...

it looks like there will be no provision in the Leap15 kernel for ReiserFS as a module any more.

...

I don't see that mentioned anywhere. It's also difficult to see why we should go to the length of changing the kernel build config to exclude a module just because we don't want to support that file system.

I agree, but that seems to be the implication if it steps through the /etc/fstab forcing you to change all the ReiserFS to something else. Why do that unless the ReiserFS module has been pulled?

No, because it is a warning and a suggestion, not a mandate. They may do that step on another release, not this one - at least as written in that announcement. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 20 September 2017 at 08:22, John Andersen <jsamyth@gmail.com> wrote:

On September 19, 2017 3:34:47 PM PDT, Anton Aylward <opensuse@antonaylward.com> wrote:

...

On 19/09/17 09:17 AM, Per Jessen wrote:

...

...

it looks like there will be no provision in the Leap15 kernel for ReiserFS as a module any more.

...

I don't see that mentioned anywhere. It's also difficult to see why we should go to the length of changing the kernel build config to exclude a module just because we don't want to support that file system.

I agree, but that seems to be the implication if it steps through the /etc/fstab forcing you to change all the ReiserFS to something else. Why do that unless the ReiserFS module has been pulled?

...

...

ReiserFS never was part of the kernel, it was always a module. Uh, gazillions of things are "part of the kernel", as modules.

Precisely. And John's argument that its old is pretty meaningless when you look at some of those others.

...

...

If you wanted it in the kernel you had to recompile it yourself. If "in the kernel" means "not as a module", you are right. Same applies to most of the other file systems and drivers.

Right now, it seems the only one hard compiled in is ext4FS. I look that the modules for my 42.2/4.2.13 and btrfs is there. It's not compiled in.

Whooosh! I never made that argument.

It was a jab at Brown's denigration of Reiserfs simply because it's mature and needs no maintenance.

Apart from the fact that ReiserFSv3 is the only filesystem I am aware of who's fsck consistently destroys data? If you store a ReiserFS v3 disk image (VM, container, etc) on a ReiserFS v3 filesystem, fsck WILL confuse your the disk image for another partition and 'restore' files from the image, corrupting the filesystem and losing data in the process. This is even possible if you do not totally wipe (not just format, but totally overwrite) a previous ReiserFS v3 partition on the same disk - Reisers' fsck otherwise will try and 'restore' data from the previous partition, overwriting the current one. This is behaviour that's abhorrent and remains unfixed. There is no sane person with any understanding of filesystems who could possibly argue that ReiserFS v3 'needs no maintenance' ; even Reiser himself felt the problems were unresolvable, which is why Reiser v4 was started to fundamentally rewrite most of the filesystem to workaround that issue. Do you really want to trust your data to a filesystem which has fundamental flaws which even the author abandoned rather than trying to fix? And sure, some die-hard volunteers are maintaining Reiser v4, but it has not been merged into the mainline Linux kernel. And it is unlikely to ever do so, because Reiser4 does not follow Linux coding standards. Do you really want to trust your data to a filesystem who's code quality is so poor there is no likelihood of it ever being merged into the Linux kernel? Whatever people may or may not do in other OBS repos is immaterial - openSUSE currently only supports filesystems that are part of the mainline Linux kernel. And that doesn't mean we fully support ALL filesystems in the kernel; v3 is still built in the mainline kernel for backwards compatibility purposes - but no one should use it. YaST no longer supports installations with it. Upgrades will force migration to a different filesystem. This is good advice - Everyone should migrate to more sensible options than Reiserv3 as soon as possible. Anything is probably a more sensible option. Anyone who ignores this might be able to handcraft a running system regardless, but they should realise they are on their own. They should expect that no bugs reported will be fixed, bug reports that mention reiserfs are likely to be closed if filesystem type is a possible factor, no compatibility can be ensured as new kernel features are enabled, breakages are most certainly possible and no effort will be made to test for them, nor fix them if they're reported. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Which fs do you prefer? Why? What is its main purpose? (ie, server, occasional use in either desktop or laptop, RAID, SSD, etc)

Sometime soon I will be forced to take the time to upgrade my system here so your input would be appreciated. By upgrade I mean to replace HDD and OS, i.e. a fresh install of something. The what is still up in the air.

I'm a fan of btrfs for all of my operating system disks (aka my root partition, containing my OS install and quite often my /home also) I use this on all of my machines - a home desktop, work desktop, work laptop, the server running https://rootco.de, and even my home router For me the killer usecase of btrfs is the snapshotting and rollback - given the majority of the machines are running some form of rolling release, often with automatic updates, it's nice having the certainty that I can rollback if something goes wrong (though I cant think of any time that I've had to rollback lately) btrfs' strong support of SSD's is a bonus given most of my machines are SSD/flash based. for my server, btrfs was also a natural choice because it's hosted with Hetzner, and their server installation tooling wouldn't let me do Raid 1 the way I wanted it. Being able to do that live on the system after the initial installation was beautiful. for my router, it doubles as a home backup appliance, where btrfs' strong support for de-duplication is proving to be very helpful. I will probably be playing with btrfs send/recieve for incremental backups in the future, but I haven't gotten around to it yet For any disks where none of the above features are beneficial, I use XFS. This is currently one or two rotating disks across all of my systems, which are just 'plain' disks for storing 'plain' data in a very plain fashion. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wednesday, 20 September 2017 16:09:56 ACST Richard Brown wrote:

[...] If you store a ReiserFS v3 disk image (VM, container, etc) on a ReiserFS v3 filesystem, fsck WILL confuse your the disk image for another partition and 'restore' files from the image, corrupting the filesystem and losing data in the process.

Straw man argument. The majority of the discussion has been around ReiserV4, not V3.

This is even possible if you do not totally wipe (not just format, but totally overwrite) a previous ReiserFS v3 partition on the same disk - Reisers' fsck otherwise will try and 'restore' data from the previous partition, overwriting the current one.

This is behaviour that's abhorrent and remains unfixed.

Even with V4?

There is no sane person with any understanding of filesystems who could possibly argue that ReiserFS v3 'needs no maintenance' ; even Reiser himself felt the problems were unresolvable, which is why Reiser v4 was started to fundamentally rewrite most of the filesystem to workaround that issue.

I don’t recall anyone arguing that V3 needed no maintenance. Besides, I could be equally critical of btrfs, having had a system end up unbootable, and unrecoverable without a complete repartition/reformat/ reinstall, not once but twice, because btrfs’s default installation parameters (as configured by the openSuSE installer) enabled snapshots to use up all available disk space, and there were no (working) tools available to fix it. THAT is ridiculous and abhorrent behaviour and anyone who entrusts their data to a filesystem that does that is equally insane (IMHO). There should at least be a warning if btrfs is installed during installation, stating, “Warning - you’ve selected btrfs which will eat up all usable space on your root partition if you don’t fix the settings, because our installation defaults are broken by design.”

[...] And sure, some die-hard volunteers are maintaining Reiser v4, but it has not been merged into the mainline Linux kernel. And it is unlikely to ever do so, because Reiser4 does not follow Linux coding standards.

That is a totally different issue.

Do you really want to trust your data to a filesystem who's code quality is so poor there is no likelihood of it ever being merged into the Linux kernel?

Falsely affirming the consequent. Not meeting one particular set of coding standards does not necessarily imply poor quality coding. -- ============================================================== Rodney Baker VK5ZTV rodney.baker@iinet.net.au CCNA #CSCO12880208 ============================================================== -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, Sep 20, 2017 at 1:25 PM, Richard Brown <RBrownCCB@opensuse.org> wrote:

https://en.opensuse.org/SDB:BTRFS#How_to_repair_a_broken.2Funmountable_btrfs...

I am yet to find anyone who has suffered any such problems when following the above documentation.

Then browse btrfs mailing list. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 20 Sep 2017 08:39:56 +0200 Richard Brown <RBrownCCB@opensuse.org> wrote:

On 20 September 2017 at 08:22, John Andersen <jsamyth@gmail.com> wrote:

...

It was a jab at Brown's denigration of Reiserfs simply because it's mature and needs no maintenance.

Apart from the fact that ReiserFSv3 is the only filesystem I am aware of who's fsck consistently destroys data?

OK, there's a bold assertion and I see there's some justification following, so lets approach this claim with an open mind ...

If you store a ReiserFS v3 disk image (VM, container, etc) on a ReiserFS v3 filesystem, fsck WILL confuse your the disk image for another partition and 'restore' files from the image, corrupting the filesystem and losing data in the process.

OK, so don't do that then! Not something I'm likely to want to do. Moving on ...

This is even possible if you do not totally wipe (not just format, but totally overwrite) a previous ReiserFS v3 partition on the same disk - Reisers' fsck otherwise will try and 'restore' data from the previous partition, overwriting the current one.

Again, don't do that then. So the claim amounts to: you can damage your data if you try to. I can do rm -rf /* as well, does that mean I should stop using GNU? Can you see why some people see your approach as overbearing?

This is behaviour that's abhorrent and remains unfixed. There is no sane person with any understanding of filesystems who could possibly argue that ReiserFS v3 'needs no maintenance' ; even Reiser himself felt the problems were unresolvable, which is why Reiser v4 was started to fundamentally rewrite most of the filesystem to workaround that issue.

Do you really want to trust your data to a filesystem which has fundamental flaws which even the author abandoned rather than trying to fix?

Are there flaws that actually affect me? Why haven't I met them yet? [snip]

And that doesn't mean we fully support ALL filesystems in the kernel; v3 is still built in the mainline kernel for backwards compatibility purposes - but no one should use it. YaST no longer supports installations with it. Upgrades will force migration to a different filesystem. This is good advice - Everyone should migrate to more sensible options than Reiserv3 as soon as possible. Anything is probably a more sensible option.

This is the piece that people are debating and where an answer from somebody knowledgeable would be useful. Is YaST going to suggest that you migrate a reiserfs filesystem to something else or is it going to refuse to continue the upgrade until you have done so? Is that all filesystems on the disks, or just those in the fstab? Is the kernel still going to support reiserfs?

Anyone who ignores this might be able to handcraft a running system regardless, but they should realise they are on their own.

They should expect that no bugs reported will be fixed, bug reports that mention reiserfs are likely to be closed if filesystem type is a possible factor, no compatibility can be ensured as new kernel features are enabled,

Surely that's an issue for the kernel team, not a distro?

breakages are most certainly possible and no effort will be made to test for them, nor fix them if they're reported.

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Please do not cc me when you post to the list, the double message is annoying. On 19/09/17 12:20 AM, Doug wrote:

I'm not sure what you are suggesting as a universal file system.

it's not about 'universal' any more than BtrFS should be universal or ext4FS should be universal. its about loosing options.

You are of course aware that Reiser is in prison, probably for the rest of his life, and not supporting the FS he devised, and I haven't heard that anyone else is, unless you are.

Yes I am aware he is imprisoned for the murder of his wife. But while he was the lead it was the organization he founded and the developers there that wrote and supported the FS. It was not a one-man-band. If you are unaware of the amount of development work that has been done on both the original ReiserFS and on the -4 since then I'm disappointed.

From your post, I do not see a recommendation for a "file system that fits all." (It would be nice if it worked not only for Linux but for FreeBSD. so that exchanging files between systems would be easy, and hardware --hard drives-- would work together.)

I don't know why you harp on the idea of one to fit all as opposed to increasing the number of options. We have a number of FS for SSDs, for example, not just a single one. I'm not sure just how much the idea of a OS independent FS at the hard drive level , this swapping between machines etc goes. There are Linux and BSD drivers for many file systems, enough to facilitate swaps. Heck, if I had reliable hardware I could read some of the DECtapes I have of UNIX V7 stuff, as well as the old 5" CP/M stuff. As it is, I do have a reader that plus in to the parallel port that lets me read/write iOmega PC100 "zip" disks under Linux just dandy. Those date back to 1994. Microsoft has tried for the "universal FS', licensing its FAT to everyone so it appears on the chips in my camera and phone, but I can read that on Linux too. Linux is about multiplicity of choice. ReiserFS isn't hard coded into the distributed kernel, its a module. There is no reason why it should not remain available as a module for the kernel in Leap 15. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-19 09:25, Daniel Bauer wrote:

Am 19.09.2017 um 05:18 schrieb Anton Aylward:

...

On 18/09/17 05:51 PM, Carlos E. R. wrote:

...

No, I stay away from btrfs. I use ext4, xfs, and some reiserfs. ...

...

openSuse 15 will no longer support ReiserFS.

That is very bad news to me. Does it mean it will be removed from the kernel upstream?

A shy questions for information: - Does this mean that soon I cannot read my external backup disks anymore? - When will that be?

openSUSE 15, he says. The next Leap release. You could read your backups using a virtual machine of an older release.

I have many external disks, some safety-backups even in another country, all with encrypted reiser-fs... Well, it would be one more reason to copy them to newer drives, something I should do anyway sometimes, probably, but it's a lot of non-productive work even involving journeys and I am moving it to the future as much as I can...

I also have some backups, but these are mostly XFS. But I have several partitions in actual use that are reiserfs. Of course they can be migrated, but performance would suffer. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

Anton Aylward wrote:

openSuse 15 will no longer support ReiserFS. This isn't a passive move, as in "no continued support, no further fixes", but is an active move. If you have any ReiserFS partitions then they will be forcibly converted to another file system whether you want that to happen or not.

That doesn't seem to quite correspond to this article/blog: https://lizards.opensuse.org/2017/08/24/yast-sprint-41/ -- Per Jessen, Zürich (11.8°C) http://www.hostsuisse.com/ - dedicated server rental in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R. wrote:

On 2017-09-19 10:18, Per Jessen wrote:

...

Anton Aylward wrote:

...

openSuse 15 will no longer support ReiserFS. This isn't a passive move, as in "no continued support, no further fixes", but is an active move. If you have any ReiserFS partitions then they will be forcibly converted to another file system whether you want that to happen or not.

That doesn't seem to quite correspond to this article/blog:

https://lizards.opensuse.org/2017/08/24/yast-sprint-41/

+++.................... Remove support for ReiserFS

The support of new installations with ReiserFS was removed from YaST in SUSE Linux Enterprise 12 and openSUSE Leap 42 but upgrades were still supported.

With SUSE Linux Enterprise 15 and Leap 15 the support of ReiserFS will be completely removed from YaST and the installer will block the upgrade of systems formatted with ReiserFS.

If some of the entries in the /etc/fstab file of the system to be upgraded is using ReiserFS, the installer will suggest to convert them to another filesystem type before migrating the system to SUSE Linux Enterprise 15 or openSUSE Leap 15.

A similar blocking error will be reported for ReiserFS root partitions. ....................++-

"Blocking the upgrade". I hope that means only if the root filesystem is reiserfs, but not if a data partition is.

Yes, that's how I read it. -- Per Jessen, Zürich (12.3°C) http://www.dns24.ch/ - your free DNS host, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

We haven't supported ReiserFS on new installations since Leap 42.1 (Nov 2015) That's almost 2 years ago, which is also 8 years after openSUSE stopped using ReiserFS by default and almost 6 years after the bankruptcy of Namesys, the only organisation backing ReiserFS. It should be no surprise to anyone that all support for ReiserFS will be withdrawn in Leap 15 (2018). It's a long overdue reflection of a simple reality that has been true for almost a decade now. That will mean no supported openSUSE version will support ReiserFS with the end of Leap 42.3 support (expected Jan 31 2019). That will be almost a decade after we started stepping away from ReiserFS - I therefore have little sympathy for any suggestion that this is somehow producing anyone any extra work. But if you still haven't gotten rid of that cursed mess of a filesystem in any of your environments, you still have over another year to find & contribute solutions if the work the YaST team have already done is insufficient for you. - Richard On 19 September 2017 at 13:22, Anton Aylward <opensuse@antonaylward.com> wrote:

On 19/09/17 05:26 AM, Carlos E. R. wrote:

...

If some of the entries in the /etc/fstab file of the system to be upgraded is using ReiserFS, the installer will suggest to convert them to another filesystem type before migrating the system to SUSE Linux Enterprise 15 or openSUSE Leap 15.

A similar blocking error will be reported for ReiserFS root partitions. ....................++-

"Blocking the upgrade". I hope that means only if the root filesystem is reiserfs, but not if a data partition is.

It does rather depend n what they mean by "similar". I take it read that ANY AND ALL ReiserFS will be "suggested" to be converted and an error reported if they are not.

-- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon?

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 19 September 2017 at 13:54, Anton Aylward <opensuse@antonaylward.com> wrote:

On 19/09/17 07:40 AM, Richard Brown wrote:

...

We haven't supported ReiserFS on new installations since Leap 42.1 (Nov 2015)

Funny, it was there when I upgraded to 42.2. I'm not on 42.3 yet.

Like I said, we haven't supported ReiserFS on new installations an upgrade is the antithesis of a new installation

...

That's almost 2 years ago, which is also 8 years after openSUSE stopped using ReiserFS by default and almost 6 years after the bankruptcy of Namesys, the only organisation backing ReiserFS.

It should be no surprise to anyone that all support for ReiserFS will be withdrawn in Leap 15 (2018). It's a long overdue reflection of a simple reality that has been true for almost a decade now.

Thank you for the emotional and irrational, though rationalized, knee-jerk, heavy handed response, Richard.

By this logic there are many file systems that continue to exist as supported modules in the current and upcoming releases that should also be removed. Take, for example, the old UNIX V7 file system -- 'sysv'. The last distribution that used that was SCO. How long have they been bankrupt?

And then there's 'minix'.

Neither sysv nor minix are supported filesystems by the openSUSE Project in any of the openSUSE distributions or our other sub-projects. Their lack of appearance in any of our tooling (YaST, kiwi, etc) are exactly the sort of thing users can expect for ReiserFS in the future. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, 19 Sep 2017 13:40:46 +0200 Richard Brown <RBrownCCB@opensuse.org> wrote:

We haven't supported ReiserFS on new installations since Leap 42.1 (Nov 2015)

That's almost 2 years ago, which is also 8 years after openSUSE stopped using ReiserFS by default and almost 6 years after the bankruptcy of Namesys, the only organisation backing ReiserFS.

https://reiser4.wiki.kernel.org/index.php/Main_Page 2016-06-06 - reiserfsprogs v3.6.25 has been released That's a bit over 1 year ago.

It should be no surprise to anyone that all support for ReiserFS will be withdrawn in Leap 15 (2018). It's a long overdue reflection of a simple reality that has been true for almost a decade now.

That will mean no supported openSUSE version will support ReiserFS with the end of Leap 42.3 support (expected Jan 31 2019).

That will be almost a decade after we started stepping away from ReiserFS - I therefore have little sympathy for any suggestion that this is somehow producing anyone any extra work.

It's going to cause me extra work, somehow transferring my main data and backups to some different filesystem. How is that done, without buying extra hardware, BTW? Being able to rely on linux to continue to support things for decades is one reason I use it. And here it isn't that linux is stopping support, it's just that my [soon-to-be-ex?] favourite distro appears to be going out of its way to try to prevent me continuing to use all of linux.

But if you still haven't gotten rid of that cursed mess of a filesystem in any of your environments, you still have over another year to find & contribute solutions if the work the YaST team have already done is insufficient for you.

All I know is that as a user, it's never given me any grief, unlike ZFS, BtrFS, XFS (and I don't use ext*). What solution is needed? All I can see is extra code in YaST to STOP things working as they were.

- Richard

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, 19 Sep 2017 11:26:00 +0200, Carlos E. R. wrote:

On 2017-09-19 10:18, Per Jessen wrote:

...

Anton Aylward wrote:

...

openSuse 15 will no longer support ReiserFS. This isn't a passive move, as in "no continued support, no further fixes", but is an active move. If you have any ReiserFS partitions then they will be forcibly converted to another file system whether you want that to happen or not.

That doesn't seem to quite correspond to this article/blog:

https://lizards.opensuse.org/2017/08/24/yast-sprint-41/

+++.................... Remove support for ReiserFS

The support of new installations with ReiserFS was removed from YaST in SUSE Linux Enterprise 12 and openSUSE Leap 42 but upgrades were still supported.

With SUSE Linux Enterprise 15 and Leap 15 the support of ReiserFS will be completely removed from YaST and the installer will block the upgrade of systems formatted with ReiserFS.

If some of the entries in the /etc/fstab file of the system to be upgraded is using ReiserFS, the installer will suggest to convert them to another filesystem type before migrating the system to SUSE Linux Enterprise 15 or openSUSE Leap 15.

A similar blocking error will be reported for ReiserFS root partitions. ....................++-

"Blocking the upgrade". I hope that means only if the root filesystem is reiserfs, but not if a data partition is.

I hope your are right but I'm not so optimistic. Why can't the install/upgrade occur on reiserfs? Especially if the file system is already there. The only reason can be that the system can not handle it anymore. Maybe not Leap 15 but the following release. Or else it would be nonsense. If it only was yast and the drivers would be kept, the file system wouldn't matter at all, would it? Istvan -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, Sep 19, 2017 at 3:03 PM, Per Jessen <per@computer.org> wrote:

Istvan Gabor wrote:

...

Why can't the install/upgrade occur on reiserfs? Especially if the file system is already there.

It seems to me it ought to be possible, just as we do today with JFS. The filesystem cannot be created by YaST, but if it's already there, it works just fine.

I would imagine that the needed kernel drivers for that file system would not be present. Isn't that ultimately what is changing? -- Roger Oberholtzer -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-19 15:06, Roger Oberholtzer wrote:

On Tue, Sep 19, 2017 at 3:03 PM, Per Jessen <per@computer.org> wrote:

...

Istvan Gabor wrote:

...

Why can't the install/upgrade occur on reiserfs? Especially if the file system is already there.

It seems to me it ought to be possible, just as we do today with JFS. The filesystem cannot be created by YaST, but if it's already there, it works just fine.

I would imagine that the needed kernel drivers for that file system would not be present. Isn't that ultimately what is changing?

I see the JFS files in the openSUSE kernel tree (and in grub2!). And the jfs.ko file. Removing them would be extra work. Do the openSUSE kernel devs go to the extra work of actually removing the reiserfs module from the kernel source tree? -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-19 13:38, Anton Aylward wrote:

On 19/09/17 04:18 AM, Per Jessen wrote:

...

Anton Aylward wrote:

...

openSuse 15 will no longer support ReiserFS. This isn't a passive move, as in "no continued support, no further fixes", but is an active move. If you have any ReiserFS partitions then they will be forcibly converted to another file system whether you want that to happen or not.

That doesn't seem to quite correspond to this article/blog:

https://lizards.opensuse.org/2017/08/24/yast-sprint-41/

The way I read it there is no alternative to conversion offered.

No. I read it as it blocks on error if it is on "/", offers change otherwise. Offers, not force. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-20 01:06, Anton Aylward wrote:

On 19/09/17 08:19 AM, Carlos E. R. wrote:

...

...

The way I read it there is no alternative to conversion offered.

...

No. I read it as it blocks on error if it is on "/", offers change otherwise. Offers, not force.

Quite correct. You are offered a change from ReiserFS to some other FS. If you refuse the offer it constitutes an error and that file system is no longer part of the new /etc/fstab since, as Mr Brown points out, support being not there means that the module is no longer there.

He may say, but the developers DO NOT say that. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

Dave Plater wrote:

I think all that is happening is openSUSE is dropping support for reiserfs (Which I fondly remember running SUSE 6.4 on recycled hard drives, even one that was clunking, using reiserfs) but I don't think that it will be maliciously removed from the kernel.

Agree completely. There is simply no reason.

That would prevent yast from converting the partitions in the first place. If you can't read a partition you can't convert it.

This conversion idea seems a pretty odd and risky thing to attempt - fiddling people's data, needing extra space, backup etc. -- Per Jessen, Zürich (10.6°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-20 08:59, Dave Plater wrote:

On 20/09/2017 08:50, Per Jessen wrote:

...

Dave Plater wrote:

...

I think all that is happening is openSUSE is dropping support for reiserfs (Which I fondly remember running SUSE 6.4 on recycled hard drives, even one that was clunking, using reiserfs) but I don't think that it will be maliciously removed from the kernel.

Agree completely.  There is simply no reason.

...

That would prevent yast from converting the partitions in the first place. If you can't read a partition you can't convert it.

This conversion idea seems a pretty odd and risky thing to attempt - fiddling people's data, needing extra space, backup etc.

It sounds like a lot of work for the yast/installation team as well, as you said it's risky, I don't know the structure of the reiser file system but it would need to be converted into a similar structured file system or it would require a lot of spare space to backup the system and then copy over to a new partition. This would be a show stopper for users without enough space if a reliable partition converter hasn't been created.

Converting live, no, I mean, on site, a reiserfs partition to... what? Reiserfs structure is a nightmare to understand and code, IMHO because it is brilliant. Write a migration tool in-house? Wow. Small files it stores where a normal filesystem stores the file name, without reserving data sectors to it. If I recall correctly, as the data sector block is not determined, it might store two files to the same sector.[1] How can you code conversion of that? Only by copying the files elsewhere. Creating a small filesystem then grow it. [1] Wikipedia explains it better. «ReiserFS stores file metadata ("stat items"), directory entries ("directory items"), inode block lists ("indirect items"), and tails of files ("direct items") in a single, combined B+ tree keyed by a universal object ID. Disk blocks allocated to nodes of the tree are "formatted internal blocks". Blocks for leaf nodes (in which items are packed end-to-end) are "formatted leaf blocks". All other blocks are "unformatted blocks" containing file contents. Directory items with too many entries or indirect items which are too long to fit into a node spill over into the right leaf neighbour. Block allocation is tracked by free space bitmaps in fixed locations.» «By contrast, ext2 and other Berkeley FFS-like file systems of that time simply used a fixed formula for computing inode locations, hence limiting the number of files they may contain.[21] Most such file systems also store directories as simple lists of entries, which makes directory lookups and updates linear time operations and degrades performance on very large directories. The single B+ tree design in ReiserFS avoids both of these problems due to better scalability properties.» https://en.wikipedia.org/wiki/ReiserFS -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 2017-09-20 13:02, Dave Plater wrote:

...

Everything is doable but the complexity of the task would increase the possibility of bugs. Anyway it seems that fstransform claims to convert between all the major filesystems.

Mmm... No wikipedia article. <http://www.linux-magazine.com/Online/Features/Converting-Filesystems-with-Fstransform> «For example, Fstransform only supports the common filesystems on Linux: ext2, ext3, ext4, ReiserFS, JFS, and XFS.» not btrfs, or it is new. «Additionally, the danger always exists with Fstransform that the conversion will dump all your data into a black hole. Ghilardi expressly assumes no liability for data loss and issues a warning to this effect in the documentation: In other words, if you lose data, it’s your problem. Thus, if you want to use Fstransform, you should create a backup copy first, preferably in the form of an image. Then, you can quickly revert to the previous state if necessary. The Clonezilla Live CD is perfect for this.» There is a requirement table that I can not copy, it is an image. «The fstransform program is just a script; the actual conversion in the background is handled by two programs: fsmove and fsremap.» «Fstransform is extremely useful, especially if you want to convert filesystems that have major technical differences – such as ReiserFS to ext4. Because Fstransform fires some pretty big guns for this job, you should take the warnings very seriously and make sure you have a usable backup. Incidentally, some filesystems come with their own conversion programs, which can be somewhat faster and safer than Fstransform.»

I dropped reiser for xfs way back when it's demise was first announced. My computer has to permanently run on a ups connected to a large battery because my mains supply never exceeds 210v (230v supply) and lives at 170v during peak demand sometimes drops below 120V. I've a 60AH deep cycle battery on the ups and a voltage tolerant power supply to keep it topped up.

Wow.

So my prime requirement for a file system is power failure tolerance and xfs seems to be the best for that. Saying that ext4 appears to be ok lately. The only problem I've had for ages is a corrupted file on a usb drive with an ntfs partition. I had to mount it in an ms VM to fix it.

I remember that when SuSE pushed ReiserFS before it was upstreamed at the kernel, one of its main "pro's" was something like power failure resilience, specially when compared to the ext2 of the time. Maybe to ext3, too. I have reduced my usage of reiserfs, migrating to XFS. But there are tasks where reiserfs 3 still outperforms any other. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On Wed, 20 Sep 2017 13:02:22 +0200 Dave Plater <dplater.list@gmail.com> wrote:

On 20/09/2017 12:36, Carlos E. R. wrote:

...

On 2017-09-20 08:59, Dave Plater wrote:

...

On 20/09/2017 08:50, Per Jessen wrote:

...

Dave Plater wrote:

...

I think all that is happening is openSUSE is dropping support for reiserfs (Which I fondly remember running SUSE 6.4 on recycled hard drives, even one that was clunking, using reiserfs) but I don't think that it will be maliciously removed from the kernel.

Agree completely.  There is simply no reason.

...

That would prevent yast from converting the partitions in the first place. If you can't read a partition you can't convert it.

This conversion idea seems a pretty odd and risky thing to attempt - fiddling people's data, needing extra space, backup etc.

It sounds like a lot of work for the yast/installation team as well, as you said it's risky, I don't know the structure of the reiser file system but it would need to be converted into a similar structured file system or it would require a lot of spare space to backup the system and then copy over to a new partition. This would be a show stopper for users without enough space if a reliable partition converter hasn't been created. Converting live, no, I mean, on site, a reiserfs partition to... what?

...

Small files it stores where a normal filesystem stores the file name, without reserving data sectors to it. If I recall correctly, as the data sector block is not determined, it might store two files to

...

Only by copying the files elsewhere. Creating a small filesystem then grow it.

...

[1] Wikipedia explains it better. «ReiserFS stores file metadata ("stat items"), directory entries ("directory items"), inode block lists ("indirect items"), and tails of files ("direct items") in a single, combined B+ tree keyed by a universal object ID. Disk blocks allocated to nodes of the

...

«By contrast, ext2 and other Berkeley FFS-like file systems of that time simply used a fixed formula for computing inode locations, hence

Reiserfs structure is a nightmare to understand and code, IMHO because it is brilliant. Write a migration tool in-house? Wow. the same sector.[1] How can you code conversion of that? tree are "formatted internal blocks". Blocks for leaf nodes (in which items are packed end-to-end) are "formatted leaf blocks". All other blocks are "unformatted blocks" containing file contents. Directory items with too many entries or indirect items which are too long to fit into a node spill over into the right leaf neighbour. Block allocation is tracked by free space bitmaps in fixed locations.» limiting the number of files they may contain.[21] Most such file systems also store directories as simple lists of entries, which makes directory lookups and updates linear time operations and degrades performance on very large directories. The single B+ tree design in ReiserFS avoids both of these problems due to better scalability properties.»

...

https://en.wikipedia.org/wiki/ReiserFS Everything is doable but the complexity of the task would increase the possibility of bugs. Anyway it seems that fstransform claims to convert between all the major filesystems. I dropped reiser for xfs way back when it's demise was first announced. My computer has to permanently run on a ups connected to a large battery because my mains supply never exceeds 210v (230v supply) and lives at 170v during peak demand sometimes drops below 120V. I've a 60AH deep cycle battery on the ups and a voltage tolerant power supply to keep it topped up. So my prime requirement for a file system is power failure tolerance and xfs seems to be the best for that. Saying that ext4 appears to be ok lately. The only problem I've had for ages is a corrupted file on a usb drive with an ntfs partition. I had to mount it in an ms VM to fix it. Dave P

Dave, please would you fix your mailer so it does quoting in a readable way? Try hitting return before typing your contribution or something. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-09-21 12:50, Dave Plater wrote:

On 20/09/2017 17:54, Dave Howorth wrote:

...

...

...

scalability properties.»

...

https://en.wikipedia.org/wiki/ReiserFS Everything is doable but the complexity of the task would increase the possibility of bugs. Anyway it seems that fstransform claims to convert between all the major filesystems. I dropped reiser for xfs way back when it's demise was first announced. My computer has to permanently run on a ups connected to a large battery because my mains supply never exceeds 210v (230v supply) and lives at 170v during peak demand sometimes drops below 120V. I've a 60AH deep cycle battery on the ups and a voltage tolerant power supply to keep it topped up. So my prime requirement for a file system is power failure tolerance and xfs seems to be the best for that. Saying that ext4 appears to be ok lately. The only problem I've had for ages is a corrupted file on a usb drive with an ntfs partition. I had to mount it in an ms VM to fix it. Dave P

Dave, please would you fix your mailer so it does quoting in a readable way? Try hitting return before typing your contribution or something. I see what you mean, it looks like the wikipedia link is mine. Dave P

Just add a blank line before writing. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

ssd is very fast in any use, but of course limited by the speed of the other computer's parts for approx 2 years I only use ssd for main disk. However on a small netbook with amd processor, I removed the ssd to set back the original drive because the processor is the probme and then no significant speed gain was achieved with the ssd I try to buy long warranty ssd's and do not care of wearing. Anyway in three years my present ssd will probably outdated :-( jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

...

I can get a brand new 1T SATA SSD for around what paid for my rotating rust 1T SATA four years ago.

Are you sure? I was buying 1TB external USB drives for $80 4 years ago. (I bought and still buy them routinely for work). I think a 1TB Samsung 960 Pro SSD today is still almost $500. That's not an entry level SSD, so maybe those are down around $100/TB? Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 10/12/2017 12:03 PM, jdd@dodin.org wrote:

what's important is to take opportunities when they come.

Exactly. There are price watcher applications for most online shopping venues. https://camelcamelcamel.com/ for Amazon for example. If you are willing to wait, you can put in a price 10 or 30 % lower than the current price and just let it do the work. These things do go on sale frequently, especially if a new model is rumored, or your source gets overstocked. You are better off shopping specific models than taking any random ssd brand that happens to go on sale. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 10/12/2017 01:06 PM, Carlos E. R. wrote:

...

On 2017-10-12 15:19, Anton Aylward wrote: I can get a brand new 1T SATA SSD for around what paid for my rotating rust 1T SATA four years ago.

I doubt it. My rotating disks were all in the 60..80€ range. The cheapest 1 GB SSD I can find is 300€. 448 for a Samsung Pro.

Read carefully Carlos. Remember he's talking about a 4 year old purchase. As another point of reference, had I held off for 1 year buying my current 512 Meg Samsung SSD 850 Pro, I could have a 1TB version today for about the same money from Crucial.com or NewEgg or several other vendors and several other manufacturers. (As low as $250 in my brief look on line for 1 TB Sata SSD). So prices are coming down. While not yet to today's commodity market prices of rotating disks, its easy to see that a 4 year old purchase was well above the prices of SSDs today. -- After all is said and done, more is said than done.

On 14/10/17 02:01 PM, jdd@dodin.org wrote:

1Tb rotating disks where well under $100 4 years ago.

Indeed they were. The discount shops would - then - sell you a reconditioned 1T for around that. I'll have to go check what the exchange rates were. Are you talking US$ or AU$? I got a reconditioned. It died the next day. YMMV.

P ices use to drop, but for time to time they grow up. I guess 500 Gb ssd are enough for system, no need for 1Tb

IIR correctly the 1T was about $20 more than the 500M.

the market may go to m2 or msata disks

Indeed; I wouldn't be surprised in the least. There's not a lot of use in having a fast storage device it your 'channel can't keep up with it. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-10-14 19:31, John Andersen wrote:

On 10/12/2017 01:06 PM, Carlos E. R. wrote:

...

...

On 2017-10-12 15:19, Anton Aylward wrote: I can get a brand new 1T SATA SSD for around what paid for my rotating rust 1T SATA four years ago.

...

I doubt it. My rotating disks were all in the 60..80€ range. The cheapest 1 GB SSD I can find is 300€. 448 for a Samsung Pro.

Read carefully Carlos. Remember he's talking about a 4 year old purchase.

I know. 4 years ago my rotating disks were about 60..80€ each. Same four years earlier, but smaller. Always about the same price over a decade or two. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)