On December 14, 2017 8:48:46 AM PST, James Knott wrote:

On 12/14/2017 11:34 AM, Richard Brown wrote:

...

IIRC the official openSUSE recommended sizing of swap is 2x RAM-size, but no larger than 2GB, unless you want Hibernation, in which case it is 2GB or 1x RAM-size, whichever is the larger

Care to provide a link for that? While it may have been the case with *nix systems years ago, I don't think it applies any more.

I recall having more swap than ram was recommended to allow hibernate without having to first repatriate swapped out stuff. I don't recall it ever being 2xRam. -- Sent from my Android phone with K-9 Mail. Please excuse my brevity. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Thu, 14 Dec 2017 21:45:54 +0100 David Haller wrote:

On Thu, 14 Dec 2017, John Andersen wrote:

...

On December 14, 2017 8:48:46 AM PST, James Knott wrote:

...

On 12/14/2017 11:34 AM, Richard Brown wrote:

...

IIRC the official openSUSE recommended sizing of swap is 2x RAM-size, but no larger than 2GB, unless you want Hibernation, in which case it is 2GB or 1x RAM-size, whichever is the larger

Care to provide a link for that? While it may have been the case with *nix systems years ago, I don't think it applies any more.

I'm curious. Hibernation is suspend to disk, yes? (I don't use it myself) And specifically, hibernation is suspend to 'swap' space, which really means suspend to paging space under linux. And there are all kinds of guidelines about how much space is sensible or reasonable to allocate as swap space. But that depends on workload. But what happens if swap space is all in use when hibernation is requested? Does the system refuse to hibernate? Or trash running process images? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday, 2017-12-14 at 16:46 -0500, James Knott wrote:

On 12/14/2017 04:38 PM, Dave Howorth wrote:

...

But what happens if swap space is all in use when hibernation is requested? Does the system refuse to hibernate? Or trash running process images?

That might depend on what's in swap. For example, if it's executables, which can easily be reloaded, then discard them to make room.

AFAIK, Linux doesn't do this. Windows does. No, now that I think, it doesn't: what it does is that it does not "swap" code, but dumps it and reloads from exe file when needed again. At least it did several versions back. The strategy saves disk space, but I don't think it is efficient speed wise. I understand that swapped apps at the time of hibernation are just left swapped out, untouched. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAloy8yYACgkQtTMYHG2NR9WG+ACcDx1lQC0UgPA8ElxRdDMXrTvk WkEAoI0rKG8m8eQNxOG/FkTaDY19ymHY =0cnz -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Thu, 14 Dec 2017 22:50:56 +0100 (CET) "Carlos E. R." wrote:

...

But what happens if swap space is all in use when hibernation is requested? Does the system refuse to hibernate? Or trash running process images?

AFAIK it leaves the swaped pages as they are (unless those processes need to run before being hibernated), and adds the hibernation image to swap. So you may need a lot of swap.

Yes, but that doesn't answer my question. What if swap is full? And however much swap you have, it might still be full. I'm hoping that there's some designed solution, which I'm simply ignorant of. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 14/12/2017 à 21:45, David Haller a écrit :

That was when RAM was measured in MB, not GB ... Actually, the recommendation was: "AT LEAST 2*RAM".

well, with my 16Gb ram, I even wonder if swap is needed at all :-)

free total used free shared buffers cached Mem: 16312176 14958532 1353644 143356 487296 12950524 -/+ buffers/cache: 1520712 14791464 Swap: 8384508 0 8384508

jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

ecommendation was: "AT LEAST 2*RAM". well, with my 16Gb ram, I even wonder if swap is needed at all :-)

free total used free shared buffers cached Mem: 16312176 14958532 1353644 143356 487296 12950524 -/+ buffers/cache: 1520712 14791464 Swap: 8384508 0 8384508

by the way hibernation works, I mean the computer stops and do not crash at start but how can I see if it simply stopped and restarted or come back from hibernation? jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 15/12/2017 à 00:36, David T-G a écrit :

jdd --

...and then jdd@dodin.org said... % ... % by the way hibernation works, I mean the computer stops and do not % crash at start

It shouldn't crash in any case :-)

some posts let me think otherwise :-(

% % but how can I see if it simply stopped and restarted or come back % from hibernation?

If the machine is restarted, then everything gets kicked off as a fresh process, right?

when testing at first, I didn't want to lose data so I stopped all the apps running. Today I made a test with application open and effectively got the apps open at restart I usually don't use hibernation, because I find the time to boot not that better and I use a multiboot machine, but this one is not multiboot. I still always fear to lose something, for example if I forget I hibernated thanks jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos, et al -- ...and then Carlos E. R. said... % % On Friday, 2017-12-15 at 12:12 +0100, jdd@dodin.org wrote: % ... % >I still always fear to lose something, for example if I forget I hibernated % % Hibernation disables the grub menu. Wait, what? NOOOO! That is exactly what I don't want :-( Do you know how to change that so that I get the menu no matter which OS instance shut down or hibernated last? TIA again & HH :-D -- David T-G See http://justpickone.org/davidtg/email/ See http://justpickone.org/davidtg/tofu.txt -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

David T-G composed on 2017-12-15 15:33 (UTC-0500):

...and then Carlos E. R. said... % % On Friday, 2017-12-15 at 12:12 +0100, jdd@dodin.org wrote: % ... % >I still always fear to lose something, for example if I forget I hibernated % % Hibernation disables the grub menu.

Wait, what? NOOOO! That is exactly what I don't want :-(

Do you know how to change that so that I get the menu no matter which OS instance shut down or hibernated last?

To put more simply what Carlos explained, hibernation as a practical matter is incompatible with multiboot. -- "Wisdom is supreme; therefore get wisdom. Whatever else you get, get wisdom." Proverbs 4:7 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 15/12/2017 à 18:59, Carlos E. R. a écrit :

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On Friday, 2017-12-15 at 12:12 +0100, jdd@dodin.org wrote:

...

some posts let me think otherwise :-(

It /might/ crash. I have not seen it in years, though. I have had problems with hibernation on some openSUSE versions.

normally it shouldn't, I guess it still can stop the computer if hibernation fails, but sure some years ago hibernation didn't work well right now I tested it on my config (16Gb ram, 8Gb swap) and it works. I was pretty sure it only needs approx half the max ram *used* really (no cache). I don't see how I could simulate a full 16Gb ram used, may be with lot of virtual machines? pretty long to test

...

...

% %  but how can I see if it simply stopped and restarted or come back %  from hibernation?

I worked on this. hibernation makes "journactl -b 0" unusable o nearly, as there is no reboot. One have to search "sleep requested", "starting hibernate" "suspended system" in logs and what is around them. curiously enough, part of the logs of the hibernation are written *after* reboot, as ascertained by the logs date, probably a buffer saved but not written in logs. "creating hibernation image" speaks of "pages". As I don't know the page size I can't know the hibernation file size. Only 14Mb remain in swap after restore. My image was 1 206 945 pages the computer is stopped on a pci error and "sleep state S4"

When using many applications at the same time, it saves me time to find them open instead of starting each one again. I prefer hibernation.

Hibernation disables the grub menu.

sure. hibernation forces the use of the last system used in my present computer (I have it for only some weeks), it's extremely fast - around 5s to shut down, and same for starting - the password prompt is the longer to do. But then booting fresh is only around 15s jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday, 2017-12-14 at 18:36 -0500, David T-G wrote:

% % but how can I see if it simply stopped and restarted or come back % from hibernation?

Reading the logs :-)

If the machine is restarted, then everything gets kicked off as a fresh process, right?

In the case of waking from hibernation, the processes all pick up right where they left off. Think of it like forking a process; the new fork didn't have to get called but instead just exists.

I get the screen saver active on return, enter my password, and see all my opened apps and files in the same point as they were. You might not notice if the desktop was empty.

If you really need to know, I'd write a flag file or similar as part of the hibernation script and then detect it as you wake up and come out of the system call. In practice, though, you basically won't know it happened :-)

I write entries in a log file (/var/log/Sesiones.log): 2017-11-08 15:04:54+01:00 - Booting the system now ================================================================================ Linux Telcontar 4.4.92-18.36-default #1 SMP Tue Oct 24 15:20:18 UTC 2017 (3f3cfaa) x86_64 x86_64 x86_64 GNU/Linux 2017-11-08 20:09:15+01:00 - Hibernating the system now... 2017-11-08 23:24:23+01:00 - Thawing the system now)... 2017-11-09 04:39:05+01:00 - Hibernating the system now... 2017-11-09 09:45:46+01:00 - Thawing the system now)... ... 2017-12-05 04:15:03+01:00 - Halting the system now =========================================== uptime: 04:15am up 26 days 13:10, 1 user, load average: 0.69, 0.68, 0.66 2017-12-05 11:46:23+01:00 - Booting the system now ================================================================================ Linux Telcontar 4.4.92-18.36-default #1 SMP Tue Oct 24 15:20:18 UTC 2017 (3f3cfaa) x86_64 x86_64 x86_64 GNU/Linux 2017-12-06 04:18:05+01:00 - Hibernating the system now... - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo0CygACgkQtTMYHG2NR9UsZQCePPX3fLf8BYfM1bZWKQsYbozG YioAn2kj0QDHTEhX8QDXdIIuFsCSVy2O =AzD1 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 14 December 2017 at 17:48, James Knott wrote:

On 12/14/2017 11:34 AM, Richard Brown wrote:

...

IIRC the official openSUSE recommended sizing of swap is 2x RAM-size, but no larger than 2GB, unless you want Hibernation, in which case it is 2GB or 1x RAM-size, whichever is the larger

Care to provide a link for that? While it may have been the case with *nix systems years ago, I don't think it applies any more.

https://www.novell.com/support/kb/doc.php?id=7010157 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 12/15/2017 03:48 AM, Richard Brown wrote:

On 14 December 2017 at 17:48, James Knott wrote:

...

On 12/14/2017 11:34 AM, Richard Brown wrote:

...

IIRC the official openSUSE recommended sizing of swap is 2x RAM-size, but no larger than 2GB, unless you want Hibernation, in which case it is 2GB or 1x RAM-size, whichever is the larger Care to provide a link for that? While it may have been the case with *nix systems years ago, I don't think it applies any more. https://www.novell.com/support/kb/doc.php?id=7010157

Well, lessee now... "Is swap space required? Strictly speaking, a swap partition is not required for SLES. However, swap space is recommended." and "How much swap space is recommended? Every server environment is different, and swap space requirements vary greatly depending upon the software environment. As a general recommendation, a swap space size of 2GB is typically sufficient. (Some large database environments may require much larger swap spaces.)" I don't see anything that says swap must be 2x memory, as some claim. I see it can be any amount according to the above, but 2 GB recommended. The thing about swap is it's virtual memory. It pretends to be memory space. This means, the more memory you have, the less swap you need. That's always been the case for as long as I can remember, on any OS. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 15/12/17 10:39 AM, Wols Lists wrote:

For most of its history, the twice ram rule has been a "needed for optimum performance", not a hard and fast "needed or it won't work" rule, but it is solidly based in historical fact.

For traditional Unix, and pre-2.4 linux, even if it wasn't a hard requirement it also had a real performance impact thanks to the optimisations required to deal with "too small" swap spaces.

*Sigh* The moment we mention 'traditional UNIX' my hackles go up. Are we talking post 4.2 Berkeley on the VAX with hardware (even if inadequate) support for paging? Are we talking pre-4.2 with pre-paging aka swap-in/swap-out? Are we talking the mess that Bell Systems Group did with SYSIII and SYSV (let's not even think about SYSIV!) UNIX? So much of this talk is of swap space but it not being used for swap, it's being used for EITHER demand paging memory overflow ==> call this 'page-out space' OR hibernation The latter DOES require AT LEAST as much disk space as memory space, plus any additional that might be needed for any page overflow. If you are NOT hibernating then you may not need page-out space. Why would you need 'page-out- space? That's the issue. I have ideas about that and they are based on experience rather than theory, but for most non-public-server applications I can see the 2G being all that's required for initialization. Traffic analysis might show you need more, but it is easy enough to add more as a partition or as a 'swap file'. Real world =/= Theory But PLEASE DIFFERENTIATE: Swapping and demand page overflow page-out space are not the same thing. If you have a virtual memory system that is 'paging', that is, it is swapping out the whole of the working set of a process to make room for the working set of another process, then you are 'thrashing'. You are in dire straits and something is wrong. It is usually an indication that your system is memory starved, and on a modern system that should not happen -- unless you are running a web browser with lots-and-lots of open tabs. As I see it, lots-and-lots of open tabs means lots-and-lots of open file descriptors in the kernel. The kernel can malloc(), it's internal tables can grow. See 'kmalloc()' and 'vmalloc()'. One such is dynamically loading kernel modules using "modprobe(8)". You might want to think, 'does the kernel free up that space when you remove one with "modprobe -r"?' And what about the other kernel tables, for example the open file and open network and the buffers needed for the network connections as they are opened? Does the kernel VM do page-out? I don't think so. So as the kernel grow the user space decreases and that does the page-out or the amount of memory for the working sets decreases and perhaps to a point where there is thrashing. So when I map my 80+ tabs of Firefox, of which only 4 are actually open, to Chromium, where there is a process for each tab and a each has an open network connection, no wonder my machine freezes! All of a sudden it is so busy thrashing it can't do any useful work. THAT is the point where I need to SWAP. Swap out whole processes, wholesale, rather than treating them as sets of pages and trying to maintain them all. But Linux doesn't swap? -- 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

* David T-G [12-15-17 12:07]: [...]

[snip]

Ironically, that is very much part of my typical profile; I run both Chrom(e|ium) and FireFox, and they're always loaded with lots of tabs in lots of windows (as multiple users, even, all on the same display) across multiple desktops (usually general comm, personal play, work, and tinkering) on a given machine, plus anything else I'm doing, and so my hotshot mobile workstation is ... adequate :-)

And I'm very interested in hibernation both because it's a laptop that goes with me and because I still, for now at least, dual-boot and would rather take the shortcut of going to sleep and then coming back to where I was than to fully shut down and then start over.

So I know that I need hibernation storage space, and I'm happy to give up 8G of disk for that. [Any thoughts on swapon-ing a file as I get ready to hibernate?] And I strongly suspect, given my abusive habits, that I'll end up needing process swap "page out" space even with 8G of RAM, and I don't mind giving up space for that *but* also don't want to give my system too much that can apparently slow it down. I just don't know where to draw the line and how to move it around later...

you could always set swap on/off via a script prior to hybernation, *and* you could set an additional swap area as a swap file smaller than the requirements for hibersquash and use it when active. or just set up two swap areas and use one for active time and the other for hibersleep via a script with swap on/off. -- (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

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Friday, 2017-12-15 at 12:05 -0500, David T-G wrote: ...

So I know that I need hibernation storage space, and I'm happy to give up 8G of disk for that. [Any thoughts on swapon-ing a file as I get ready to hibernate?] And I strongly suspect, given my abusive habits, that I'll end up needing process swap "page out" space even with 8G of RAM, and I don't mind giving up space for that *but* also don't want to give my system too much that can apparently slow it down. I just don't know where to draw the line and how to move it around later...

Some misconception here :-) The system doesn't slow because it is using swap, it slows because it needs more ram, and there isn't. Example: Suppose that the system sends to swap the part of processes that are only used once during initialization. Those parts are never used again. The result is used swap, and more free ram! The machine actually becomes faster. This situation happens when using hibernation: on return some swap remains in use and does not disappear, even if the system normally doesn't swap during usage. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo0EWEACgkQtTMYHG2NR9W4KgCdHEFCtYb762XhoGdEq8cau2Rt lloAnifO30ooc0WkSqN0X4oodhfgr7Qn =C/RG -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward wrote:

So when I map my 80+ tabs of Firefox, of which only 4 are actually open, to Chromium, where there is a process for each tab and a each has an open network connection,

I think Firefox, by default, does the same. I had a bug open on some rendering issue - a work-around was to disable the per-tab process. AFAIR. -- Per Jessen, Zürich (4.4°C) http://www.hostsuisse.com/ - virtual servers, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 14/12/2017 à 17:34, Richard Brown a écrit :

IIRC the official openSUSE recommended sizing of swap is 2x RAM-size, but no larger than 2GB, unless you want Hibernation, in which case it is 2GB or 1x RAM-size, whichever is the larger

interesting. Do you have a link to this "official" recommendation? (the question come often) thanks jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos, et al -- ...and then Carlos E. R. said... % % On Thursday, 2017-12-14 at 16:11 -0500, David T-G wrote: % ... % >What does it mean when the "expand swap?" check box is not active, % >and what do I need to do to allow hibernation (or hybrid sleep, even, % >if that also allows me to pull all power and park it for a month) as % >well as suspension? % % If you are installing a laptop and intend to hibernate it at some % time, then tick the box. If you don't like what it proposes, change % it manually. I have never used that tick box. But ... it's not active and I don't have the choice of ticking it. % % Yes, hybrid will allow you to park for months. However, as the % battery is active during the suspend phase, powering the RAM till it % dies in, I don't know, three days?, well after that the battery will % remain discharged, which is not good. Yeah; that's why I'd really prefer hibernate. % % - -- Cheers, % Carlos E. R. Thanks again & HH :-D -- David T-G See http://justpickone.org/davidtg/email/ See http://justpickone.org/davidtg/tofu.txt -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Thu, 14 Dec 2017 17:50:25 -0500 David T-G wrote:

Carlos, et al --

...and then Carlos E. R. said... % % On Thursday, 2017-12-14 at 11:17 -0500, David T-G wrote: % ... % >How much swap do I need if I'm planning to suspend to disk (aka hibernate % >as compared to sleep, but I forget the s2/s3/... stages)? % % Well, you need as much as you need... :-)

Gee, thanks :-)

% % There can't be an exact reccomendation. You initial choice of 17G is % fine (it is what I have, more or less). Only experience with your % actual usage pattern will tell. [snip]

Well, dang. I guess I share Dave Howorth's position in that I'd like to know some real guidelines. Who knew that Window's hiberfil.sys for dumping RAM to disk alongside the variable-size pagefile.sys for its equivalent of swapping would be so useful?

Soooooo... I think I hear that swap space is used to write the current active memory for hibernation as well as to swap out process data for CPU time sharing, and I think I hear that there are is no firm way to ensure that one always has enough "hibernation" space free. Is that accurate?

/me wonders if creating an 8G file in OS space and using swapon to turn it on right before hibernation (and then off again upon waking up) is a reasonable plan...

I'm surprised nobody has mentioned the OOM killer in this discussion. Does it get called to free up swap for hibernation?

Thanks again & HH

:-D

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

On Friday, 2017-12-15 at 00:52 -0000, Dave Howorth wrote:

...

On Thu, 14 Dec 2017 17:50:25 -0500 David T-G <> wrote:

...

Carlos, et al --

...

...

I'm surprised nobody has mentioned the OOM killer in this discussion. Does it get called to free up swap for hibernation?

Eum... :-?

I don't know if I see the relati... Ah, I do. No, it doesn't. It simply bails out as clean as possible, says nothing, and you have to decide what to do.

Well, it says what happened in the log. Which, if the screensaver lock kicked in, you don't see. It doesn't even beep.

You can not simply close the lid and put the laptop in the bag without looking. Do wait for it to actually suspend or hibernate.

-- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) Eum? That's what I've been doing all the time, for years now. Close the lid,

Op vrijdag 15 december 2017 18:32:03 CET schreef Carlos E. R.: put the laptop in the backpack and off I go. Never failed once. To me this sounds like checking whether the light goes off after closing the refrigerator door. 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

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Friday, 2017-12-15 at 18:50 +0100, Per Jessen wrote:

Carlos E. R. wrote:

...

You can not simply close the lid and put the laptop in the bag without looking. Do wait for it to actually suspend or hibernate.

I always just close the lid - why should I wait?

I see sometimes hibernation fail in my desktop, and it aborts. Some of those times there is some error that I have to solve, and others I see nothing and on retry it simply works. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo0EisACgkQtTMYHG2NR9XiAQCfT3SUg39XmpcLAkHtW7EM08Mg N0kAn2WUk2WDuJY0t5n1RFssmGX/pDw0 =j4yD -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos, et al -- ...and then Carlos E. R. said... % % On Thursday, 2017-12-14 at 17:50 -0500, David T-G wrote: % % >...and then Carlos E. R. said... % >% % >% On Thursday, 2017-12-14 at 11:17 -0500, David T-G wrote: % >% % >... % >% >How much swap do I need if I'm planning to suspend to disk (aka hibernate % >% >as compared to sleep, but I forget the s2/s3/... stages)? % >% % >% Well, you need as much as you need... :-) % > % >Gee, thanks :-) % % I know, I know... :-) % % It is a recursive statement. Linux has a few. This is mine ;-) *grin* % % >Well, dang. I guess I share Dave Howorth's position in that I'd like % >to know some real guidelines. Who knew that Window's hiberfil.sys for % >dumping RAM to disk alongside the variable-size pagefile.sys for its % >equivalent of swapping would be so useful? % % My guideline: with 8 GiB of RAM, make it 10..20 GiB of swap: % possibly 16, which is double the RAM :-p. Yep. And I'm at just over 2x, so I'm fine there ... AFAIK. % % If you don't intend to ever hibernate, 6 should suffice. Many people % could go along with nothing - you see, it does depend on what is % your actual workload. % % 2 GiB is too little: I'm using 3.3 this instant. Agreed. I'm likely to be that guy, too. % % Yesterday I was using "shotwell", a photo program. It alone had more % than 2 GB in swap, and my total was 6, IIRC. Not a lot of photo, but I'll definitely be playing with ImageMagick as well as tinkering with OpenShot. [Oh, remind me to go back through the video editor thread and ask if there's anything that will read/write iMovie project files.] % % Disk space is cheap, nowdays, so make it big. I just looked at mine: % 25 GiB in SSD. I made it real big because SSD likes emtpy space to % play with, although I don't know if I did it right for sure. Yeah; I'm inclined to agree. For now I'm spinning 1TB; eventually I hope to move to a 256B SSD (or whatever we will call an M.2 "drive" at that time) and reserve spinning for slow-is-ok and long-term stuff. In any case, though, I can certainly afford swap. % % >Soooooo... I think I hear that swap space is used to write the current % >active memory for hibernation as well as to swap out process data for CPU % >time sharing, and I think I hear that there are is no firm way to ensure % >that one always has enough "hibernation" space free. Is that accurate? % % Yes, although the kernel should know and abort nicely if not % possible to hibernate. I haven't been in that situation for years, % so I can't say. I sure hope so -- but I also don't want to get caught unable to hibernate because I'm too busy, because that's my steady state! :-) % % >/me wonders if creating an 8G file in OS space and using swapon to % >turn it on right before hibernation (and then off again upon waking up) % >is a reasonable plan... % % Nope. Doesn't seem worth the hasle to me. Takes time to prepare. No time; I'd prepare the sparse file once and then just turn it on and off as part of the hibernation script. What, me do manual work?!? :-) % Also the kernel needs to known at boot time where is the hibernation % image. Hmmm... That's a good point. I wonder how it knows whether or not to go there, because it would always be the same thing... HH :-D -- David T-G See http://justpickone.org/davidtg/email/ See http://justpickone.org/davidtg/tofu.txt -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 15 December 2017 at 18:24, Carlos E. R. wrote:

My guideline: with 8 GiB of RAM, make it 10..20 GiB of swap: possibly 16, which is double the RAM :-p.

If you don't intend to ever hibernate, 6 should suffice. Many people could go along with nothing - you see, it does depend on what is your actual workload.

2 GiB is too little: I'm using 3.3 this instant.

Yesterday I was using "shotwell", a photo program. It alone had more than 2 GB in swap, and my total was 6, IIRC.

Disk space is cheap, nowdays, so make it big. I just looked at mine: 25 GiB in SSD. I made it real big because SSD likes emtpy space to play with, although I don't know if I did it right for sure.

Carlos This is bad advice As I already provided in my earlier posts on this thread The more swap space you allocate, the more the kernel will try to use it Of course if you have a swap space larger than 6GB, you will be using more than 2GB The 2.4 and later swap algorithms will always try to use a good chunk of the available swap space But this can lower your overall system performance generally This WILL lower your overall system performance in the case of a memory leak This WILL cause delays of HOURS during a legitimate OOM condition, because your system will take HOURS paging to disk before killing applications during the OOM killer You are giving _bad_advice_ to suggest people use 25GB SWAP space in all but the most extreme conditions (eg. large, on disk, not in-memory, databases, of a size 1TB or larger, with something like 64GB of tables in memory at one time, for example) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 2017-12-15 22:54, Roger Price wrote:

On Fri, 15 Dec 2017, Richard Brown wrote:

...

As I already provided in my earlier posts on this thread The more swap space you allocate, the more the kernel will try to use it

Of course if you have a swap space larger than 6GB, you will be using more than 2GB

Is this true?  On the 42.3 machine on which I am typing this, command free reports:               total       used       free     shared    buffers     cached  Mem:      32931508   10458908   22472600     144576     648768    4378808  -/+ buffers/cache:    5431332   27500176  Swap:     66297844          0   66297844

32 GByte of memory, 63 Gbyte swap.  0 swap in use after 9 days.

That's right. The kernel is rather more clever than what Richard thinks :-)

...

The 2.4 and later swap algorithms will always try to use a good chunk of the available swap space

This is not happening for me.  Roger

QED :-) -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On 15/12/17 04:22 PM, Richard Brown wrote:

The more swap space you allocate, the more the kernel will try to use it

Experimentally, I see no evidence for that. What is there about the algorithms that would make it so? As far as I can tell, propensity to page out is a function of available memory and kernel parameters, such a swappiness, the degree of overcommit allowed, the freepages high, low and min setting. And how much gets used emerges from the kswapd parameters. All these are measures of the memory and its use by processes. Perhaps you are getting confused with the algorithm that determines if swap space is recovered. At first it is not; it is only when there is 'pressure' on the finding of swap space that the recovery algorithms are triggered. if so, then your wording is very misleading. If you have a 2G swap compression it will be triggered when you have consumed, perhaps 1.5G. Not that you have 1.5G swapped out right now, but over the lifetime of the machine running there have been that many pages swapped out at some time. Unlike real swap systems such as V7 UNIX there is not a 1:1 correspondence between the system/process memory and the page-out space, it is a first-come/first/served type of thing. (It is irrelevant in this example of the 'gap' is 3/4 or 0.5G) So it you have a 6G swap space the kernel will keep using it until the higher threshold. You can view this as a better situation since there isn't the pressure on the kernel so earnestly. There is nothing about the increased size of the swap that ENCOURAGES the kernel to use it. The extra size just means that the kernel doesn't have to run the swap space recovery/compression algorithm so often. Of course if you have adequate memory and never page-out, then the size of the swap is irrelevant. Which means that if you have additional REAL MEMORY you *DO* encourage the kernel to use it -- the *memory* that is. -- 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

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Friday, 2017-12-15 at 18:21 -0500, Anton Aylward wrote:

On 15/12/17 04:22 PM, Richard Brown wrote:

...

The more swap space you allocate, the more the kernel will try to use it

Experimentally, I see no evidence for that. What is there about the algorithms that would make it so? As far as I can tell, propensity to page out is a function of available memory and kernel parameters, such a swappiness, the degree of overcommit allowed, the freepages high, low and min setting. And how much gets used emerges from the kswapd parameters.

All these are measures of the memory and its use by processes.

Perhaps you are getting confused with the algorithm that determines if swap space is recovered. At first it is not; it is only when there is 'pressure' on the finding of swap space that the recovery algorithms are triggered.

if so, then your wording is very misleading.

If you have a 2G swap compression it will be triggered when you have consumed, perhaps 1.5G. Not that you have 1.5G swapped out right now, but over the lifetime of the machine running there have been that many pages swapped out at some time. Unlike real swap systems such as V7 UNIX there is not a 1:1 correspondence between the system/process memory and the page-out space, it is a first-come/first/served type of thing. (It is irrelevant in this example of the 'gap' is 3/4 or 0.5G) So it you have a 6G swap space the kernel will keep using it until the higher threshold. You can view this as a better situation since there isn't the pressure on the kernel so earnestly.

There is nothing about the increased size of the swap that ENCOURAGES the kernel to use it. The extra size just means that the kernel doesn't have to run the swap space recovery/compression algorithm so often.

Of course if you have adequate memory and never page-out, then the size of the swap is irrelevant. Which means that if you have additional REAL MEMORY you *DO* encourage the kernel to use it -- the *memory* that is.

Right. I see the 2 gig recommendation as the strategy when in fact you do not want swap to be used. Just as a escape valve "just in case". As the minimal size to set to make the kernel not complain. So create a minimal swap, and if it ever gets used, buy more ram. In a business server scenario, I see needing swap as a bad thing. Get more ram instead, there is money involved. The link is not a recommendation for openSUSE, but for SLE. For business servers. Not the case for people on a budget. And nowdays, with the speed of NVMe "disks", I envision more swap. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo1DD0ACgkQtTMYHG2NR9XDwwCfcK4j2n3M9GJJ3AZn7MhNFvBI L/wAn0VNk0IV2VyUHaaEzkmCxhPURwb5 =8jp+ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Saturday, 2017-12-16 at 07:49 -0500, Anton Aylward wrote:

On 16/12/17 07:06 AM, Carlos E. R. wrote:

...

In a business server scenario, I see needing swap as a bad thing. Get more ram instead, there is money involved. The link is not a recommendation for openSUSE, but for SLE. For business servers.

And, lets face it, a business server is not going to need to hibernate; its a 7/24 device!

There is a use case for swap on servers, though: when mains goes out and the battery watcher daemon kicks in: hibernate the machine in emergency, in order to restore without losing anything. Al services up, nothing lost. No admin time needed, either, except to switch on.

...

Not the case for people on a budget.

Once the budget runs out, no more RAM. And before the budget runs out it is liable to go toward more 'disk'.

Right. Just what I did recently: not "more disk", but "faster disk". I plugged in an SSD. Not NVMe, I don't have PCIe, it seems. Old board. The difference in swap speed in Leap is surprising. Otherwise, the 6 GiB of used swap I had the other day would have been horrible. I simply did not notice till I looked. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo1GwMACgkQtTMYHG2NR9WC5QCff7M9eWHH4OkbU35gjZmo23Gb 6fQAniILyM0vxJP9615qVpdKcxX7PwPJ =Tt+k -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Saturday, 2017-12-16 at 14:21 +0100, Per Jessen wrote:

Carlos E. R. wrote:

...

On Saturday, 2017-12-16 at 07:49 -0500, Anton Aylward wrote:

...

...

And, lets face it, a business server is not going to need to hibernate; its a 7/24 device!

There is a use case for swap on servers, though: when mains goes out and the battery watcher daemon kicks in: hibernate the machine in emergency, in order to restore without losing anything. Al services up, nothing lost. No admin time needed, either, except to switch on.

AFAIK, our UPS monitoring daemon (apcupsd) has no such option (hibernate instead of shutdown). I guess it could be scripted.

I don't remember on which of the two daemons I saw the option mentioned. The safe thing would be attempt to hibernate, and if it fails, poweroff. But laptops do that by default, I think. At least mine does. Different daemon, of course, but same basic idea. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo1H0UACgkQtTMYHG2NR9UOYgCfd8vioJVs5zjc0Z2229Yi20i8 6QYAoIzooMGcPpV+Q/2HzFyIemQK2Oqm =u745 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Andrei Borzenkov wrote:

16.12.2017 16:21, Per Jessen пишет:

...

Carlos E. R. wrote:

...

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On Saturday, 2017-12-16 at 07:49 -0500, Anton Aylward wrote:

...

On 16/12/17 07:06 AM, Carlos E. R. wrote:

...

In a business server scenario, I see needing swap as a bad thing. Get more ram instead, there is money involved. The link is not a recommendation for openSUSE, but for SLE. For business servers.

And, lets face it, a business server is not going to need to hibernate; its a 7/24 device!

There is a use case for swap on servers, though: when mains goes out and the battery watcher daemon kicks in: hibernate the machine in emergency, in order to restore without losing anything. Al services up, nothing lost. No admin time needed, either, except to switch on.

AFAIK, our UPS monitoring daemon (apcupsd) has no such option (hibernate instead of shutdown). I guess it could be scripted.

UPS monitoring program just calls script on event, and this script can do anything. Except hibernation has exactly zero (or even below) value in this case.

Yes, that was my first thought too. Currently, when there is about 10mins left on the UPS and we (for some reason) haven't switched to the generator, everything is told to shut down. -- Per Jessen, Zürich (2.2°C) http://www.hostsuisse.com/ - virtual servers, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 16/12/2017 à 22:04, Wol's lists a écrit :

Many budget machines have only two slots, and if it's a DDR3 machine the max is likely to be (relatively) low ...

don't forget people with budget problems have most of the time limited needs. my travel laptop is perfectly friendly with 4Gb ram and a 25€ (second hand) msata 60Bg ssd (and the original 350 Gb hdd with windows), and I buy it 80€ second hand last year. my main one is much more powerfull because I make video editing, but still high end of 3 years ago, paid half the price -600€- with i7, 1Gb ram and 256Gb ssd not every Linux application can really use 4+4 cores and so many ram jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Wol's lists composed on 2017-12-16 21:04 (UTC):

Less true nowadays, but for people on a budget, upgrading ram can easily require replacing the computer ...

I had to pay more for 8GB DDR4 RAM for the B250 motherboard I bought last week than I did 27 months ago for 16GB DDR3 RAM for the B85 motherboard then. :-( -- "Wisdom is supreme; therefore get wisdom. Whatever else you get, get wisdom." Proverbs 4:7 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 14/12/17 21:44, Carlos E. R. wrote:

Some people say "2 times RAM", but that comes, IMO, from the Windows times when that was the maximum. Others disagree.

?Anton gave you the linux swap algorithm, which clearly requires twice ram ... I gave you a concrete example of a bunch of linux kernels which would crash if they didn't have twice ram ... And you still think it's a Windows thing that has nothing to do with linux!!! ??? Please read the following link: http://lwn.net/2001/0607/kernel.php3 Especially, right near the bottom, "About that swapping problem". (At that point, there was a major rewrite of the swap system, so things may (probably have) changed since then, but "twice ram" is very much a unix/linux thing! afaik it's *never* been a windows thing, because amongst other things windows doesn't swap.) Oh - and in response to the OP's request of how much he needs, the reality is that only he can work that out ... he certainly needs SPARE swap to be greater or equal to ram if he wants to be able to suspend to disk ... however much swap he is actually using will figure in the equation ... Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

* Patrick Shanahan [12-14-17 19:15]:

* Wols Lists [12-14-17 18:06]:

...

On 14/12/17 21:44, Carlos E. R. wrote:

...

Some people say "2 times RAM", but that comes, IMO, from the Windows times when that was the maximum. Others disagree.

?Anton gave you the linux swap algorithm, which clearly requires twice ram ...

I gave you a concrete example of a bunch of linux kernels which would crash if they didn't have twice ram ...

And you still think it's a Windows thing that has nothing to do with linux!!! ???

Please read the following link:

http://lwn.net/2001/0607/kernel.php3

Especially, right near the bottom, "About that swapping problem".

(At that point, there was a major rewrite of the swap system, so things may (probably have) changed since then, but "twice ram" is very much a unix/linux thing! afaik it's *never* been a windows thing, because amongst other things windows doesn't swap.)

Oh - and in response to the OP's request of how much he needs, the reality is that only he can work that out ... he certainly needs SPARE swap to be greater or equal to ram if he wants to be able to suspend to disk ... however much swap he is actually using will figure in the equation ...

I have 36g ram and *no* swap and do not suspend/hibernate. I do not have kernel crashes or other performance problems. I run Tw and do intensive graphic work, photographs.

free -m total used free shared buff/cache available Mem: 36192 6377 771 249 29043 29889 Swap: 0 0 0

-- (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

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Friday, 2017-12-15 at 09:57 +0100, Richard Brown wrote:

On 14 December 2017 at 22:44, Carlos E. R. <> wrote:

...

...

How much swap do I need if I'm planning to suspend to disk (aka hibernate as compared to sleep, but I forget the s2/s3/... stages)?

Well, you need as much as you need... :-)

There can't be an exact reccomendation. You initial choice of 17G is fine (it is what I have, more or less). Only experience with your actual usage pattern will tell.

For hibernation you need the same ammount of used RAM + the ammount of already used swap at the instant of hibernation. However, the hibernation image can be compressed, perhaps half.

This instant I'm using 6.1 GB of swa, and a bit under 5GB of RAM, so I might need up to 11GB to hibernate.

Some people say "2 times RAM", but that comes, IMO, from the Windows times when that was the maximum. Others disagree.

I had a Linux system with 20 times at least. And yes, it was actually needed.

Please avoid giving recommendations that imply that more swap is somehow better

I didn't say that.

In other words, for peak performance you want your swap to be _as small as it needs to be_, but still enough to give the kernel the space it requires to page out unused portions of RAM to disk to give it more space to allocate RAM for more useful data.

And this is the key paragraph, in line with my actual recommendation: set up as much as you actually need.

For small systems (4GB of RAM or less) the old formulae of Swap = 2x RAM are a fair approximation of 'small enough as it needs to be', but in an era where 8GB is normal and 64GB is not extreme, the simple fact is those old recommendations do not hold true.

This is where the arbitrary "2GB is normally enough" recommendation comes from.

_normally_ I'm actually using normally 2..3 GB, and yesterday I was using 6. With your recommendation, my system would crash.

Hibernation is the exception, as you need to have enough to store the system RAM to disk. However it's worth noting that the hibernation process will COMPRESS that RAM image, so sizing your swap to equal your RAM size is actually going to oversize the swap somewhat even for that usecase..but better safe than sorry considering the possibly variable compressability of the data in RAM.

I know. Which is why I always set up somewhat more than RAM.

On large systems you do however then have the compromise of slowing down general memory performance when hibernation is enabled due to a large swap space which you only needed for hibernation

On the other hand, with large swap space, in the OOM situation the operator has time to notice and kill the culprit himself. If the system becomes slow in the large RAM and SWAP situation, as your old paper describes, the solution is the kernel be aware of the situation and act appropriately. However, nobody has a 1 TB laptop machine which he wishes to hibernate. Non issue. The OP has 8 GiB, just like me.

I suppose it would be nice if we had some mechanism for hibernating to disk somewhere other than swap for that reason.

Maybe. Well, Windows does that. - -- Cheers, Carlos E. R. (from openSUSE 42.2 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlo0CdQACgkQtTMYHG2NR9X6ZQCggcE7Sv2CAWA8ccMzPFObmymC u2QAoJUojFVluODUyHbfI85agwnc2wG/ =J9vZ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org