[opensuse] howto setup a NOT slower than Windows file server ?
Hi, I'm not shooting for an enterprise grade fileserver, mind you. I'd like a save and reasonably fast box for a couple of windows clients. I know SUSE for years and therefore I grabbed a 11.1 DVD and sat up a PC with an SEMPRON 3000, 2GB RAM, a 160 + 300 GB Samsung SATAs that should deliver enough to saturate the 100 MBit-LAN at any time. It's just, that windows <--> windows transfers tend to fill a 100MBit-LAN nicely with around 10MB/s. A SUSE or Knoppix on the same machine will yield only about 6-7MB/s with SAMBA and EXT3 and default settings. OK, I'm a wee bit unfair since SUSE does this on a RAID1. With EXT3 on an unRAIDed partition it pulls around 7-8. SUSE on an unRAIDed XFS gets 10MBs writing with FTP but only 8-9 with SAMBA. On EXT3 FTP wasn't that much faster than SAMBA. On the other hand I read in wonderfully-wise-web that XFS might be faster than EXT3 but it loses stuff now and then. I looked on Gnomes system monitor while pushing cd ISOs through the wire on empty file systems on the "server". Obviously the CPU shows the efford but is by far not maxed out and RAM shouldn't be an issue with 2GB either. Depending on the filesystem I can watch the LAN throughput drop down in optically equal distances. It seems to transfer X MBs then pauses. There were just a couple of segments per ISO. I figure it must be at least 100MB each. E.g. with EXT3 and SAMBA's default settings. There are some segments that wobble around 6.5 MB then every 3cm on the monitor ;) it peaks down to 500K/s and accelerates again. My guess is there is something written to disk while the transfer gets stopped until the writing is done. How would I get a fileserver that's performance can compete with plain Windows? Is there a howto for this? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Andreas pecked at the keyboard and wrote:
Hi,
I'm not shooting for an enterprise grade fileserver, mind you. I'd like a save and reasonably fast box for a couple of windows clients. I know SUSE for years and therefore I grabbed a 11.1 DVD and sat up a PC with an SEMPRON 3000, 2GB RAM, a 160 + 300 GB Samsung SATAs that should deliver enough to saturate the 100 MBit-LAN at any time.
It's just, that windows <--> windows transfers tend to fill a 100MBit-LAN nicely with around 10MB/s.
A SUSE or Knoppix on the same machine will yield only about 6-7MB/s with SAMBA and EXT3 and default settings.
OK, I'm a wee bit unfair since SUSE does this on a RAID1. With EXT3 on an unRAIDed partition it pulls around 7-8.
SUSE on an unRAIDed XFS gets 10MBs writing with FTP but only 8-9 with SAMBA. On EXT3 FTP wasn't that much faster than SAMBA. On the other hand I read in wonderfully-wise-web that XFS might be faster than EXT3 but it loses stuff now and then.
I have used XFS exclusively for years without any loss. -- Ken Schneider SuSe since Version 5.2, June 1998 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Ken Schneider - openSUSE escribió:
I have used XFS exclusively for years without any loss.
It may have zero'ed files on power failure, I have seen that behaviour many times, dunno if it changed over the time.. -- "If this is the best God can do, I am not impressed" -George Carlin (1937-2008) Cristian Rodríguez R. Software Developer Platform/OpenSUSE - Core Services SUSE LINUX Products GmbH Research & Development http://www.opensuse.org/
In <4A01B6FF.8000201@suse.de>, Cristian Rodríguez wrote:
Ken Schneider - openSUSE escribió:
I have used XFS exclusively for years without any loss. It may have zero'ed files on power failure, I have seen that behaviour many times, dunno if it changed over the time.
ext3 will do that too, sometimes. It's just less likely since it would sync to disk every 5 seconds and xfs doesn't. There no reason you can't write a script that calls sync every 5 seconds, if you want that syncing behavior with whatever filesystem you are using. -- Boyd Stephen Smith Jr. ,= ,-_-. =. bss@iguanasuicide.net ((_/)o o(\_)) ICQ: 514984 YM/AIM: DaTwinkDaddy `-'(. .)`-' http://iguanasuicide.net/ \_/
On Wednesday 06 May 2009 19:32, Boyd Stephen Smith Jr. wrote:
In <4A01B6FF.8000201@suse.de>, Cristian RodrÃguez wrote:
Ken Schneider - openSUSE escribió:
I have used XFS exclusively for years without any loss.
It may have zero'ed files on power failure, I have seen that behaviour many times, dunno if it changed over the time.
ext3 will do that too, sometimes. It's just less likely since it would sync to disk every 5 seconds and xfs doesn't.
There no reason you can't write a script that calls sync every 5 seconds, if you want that syncing behavior with whatever filesystem you are using.
Thats why you use a UPS on a file server! Then XFS corruption on powerloss is a moot point... -- /Rikard Johnels
Rikard Johnels wrote:
Thats why you use a UPS on a file server! Then XFS corruption on powerloss is a moot point... It might be until you discover that your batteries don't hold a charge any more as I discovered this last week when we lost power. ZFS (I am using Solaris) hasn't given me any trouble so I haven't rushed out to replace the batteries just yet.
I see it really pays to read this list. I am a bit behind current tech events and did not know about the Seagate bug. I have Seagate drives but they are older and perhaps not affected by this bug. Damon Register -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, May 7, 2009 at 8:12 AM, Damon Register <damon.w.register@lmco.com> wrote:
I see it really pays to read this list. I am a bit behind current tech events and did not know about the Seagate bug. I have Seagate drives but they are older and perhaps not affected by this bug.
You can check their website and input your serial numbers to see if you have an affected drive: http://seagate.custkb.com/seagate/crm/selfservice/search.jsp?DocId=207931 And click on the check by serial number link(javascript) I THINK that's the right place to check. Been over a month since I had my problem. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Wednesday, 2009-05-06 at 12:12 -0400, Cristian Rodríguez wrote:
Ken Schneider - openSUSE escribió:
I have used XFS exclusively for years without any loss.
It may have zero'ed files on power failure, I have seen that behaviour many times, dunno if it changed over the time..
It has been argued that it is a kernel design flaw (in order to be poxix compliant), when doing renaming-moving-deleting.old file thing in one op, each operation is not guaranteed to be complete unless you do a sync in between. XFS is much affected by this, ext4 could be. But it is not really a fault of those filesystems. I had a link with a lively discussion of this... but I don't remember where exactly. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkoDasgACgkQtTMYHG2NR9WU5gCbBgwlKx0jDybn5BJkUxGRmjf6 5WcAoJi9cUmipJN4c84+zVJnlL5ZpFS/ =nGuq -----END PGP SIGNATURE-----
Andreas wrote:
Hi,
I'm not shooting for an enterprise grade fileserver, mind you. I'd like a save and reasonably fast box for a couple of windows clients. I know SUSE for years and therefore I grabbed a 11.1 DVD and sat up a PC with an SEMPRON 3000, 2GB RAM, a 160 + 300 GB Samsung SATAs that should deliver enough to saturate the 100 MBit-LAN at any time.
<snip> Andreas, For a 'file server', it doesn't take much hardware to make a plenty good file server. I have an old Athlon tbird 800 with 1G of ram serving a dozen boxes at the office as a file server and it does just fine. The key is a reasonably fast drive in the box. This old box doesnt' even support SATA, so it's just running an old IBM Deskstar 7200 rpm drive and it mows the grass just fine. It has been serving files via samba since the samba 1.x days. I have another file server based on an Athlon 2.8 GHz chip with 2 SATA 300 drives in raid1 using dmraid as well. I have yet another file server based on a Phenom 9850 with another 4 SATA 300 drives in raid1 (2 sets). To tell you the truth, I don't really notice a performance difference between any of them for a file server. Not for a webserver, speed makes a LOT of difference... And sure, it doesn't matter what size box you have, pulling a large transfer or backup over a 100TX network will saturate/but not kill any subnet. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Wednesday, 2009-05-06 at 17:21 +0200, Andreas wrote:
On the other hand I read in wonderfully-wise-web that XFS might be faster than EXT3 but it loses stuff now and then.
Only if you power it off suddenly.
I looked on Gnomes system monitor while pushing cd ISOs through the wire on empty file systems on the "server". Obviously the CPU shows the efford but is by far not maxed out and RAM shouldn't be an issue with 2GB either.
The CPU should not see the "effort", there is almost no effort needed for reading about 10 MB/s. My old computer does that when burning a DVD. You could have some app interfering. You could try monitoring the disk usage with gkrelmn. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkoCHV8ACgkQtTMYHG2NR9WzYACghQVilffzDYdLjd5yQ1Rhcsbx bN4AnRiG568YBSbEJn7ey4AAHGej4nLB =9PFA -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wednesday May 6 2009, Carlos E. R. wrote:
...
You could try monitoring the disk usage with gkrelmn.
I like GKrellM for ongoing monitoring and have it running all the time on both my systems. The next step up in fanciness is KDE System Guard. It is much more flexible than GKrellM. Note that both GKrellM and KDE System Guard can display information from remote systems.
-- Cheers, Carlos E. R.
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
* Randall R Schulz <rschulz@sonic.net> [05-06-09 19:47]:
I like GKrellM for ongoing monitoring and have it running all the time on both my systems.
The next step up in fanciness is KDE System Guard. It is much more flexible than GKrellM.
Note that both GKrellM and KDE System Guard can display information from remote systems.
Have a good look a conky, but do not see possibility of monitoring a remote host.... -- Patrick Shanahan Plainfield, Indiana, USA HOG # US1244711 http://wahoo.no-ip.org Photo Album: http://wahoo.no-ip.org/gallery2 Registered Linux User #207535 @ http://counter.li.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wed, May 6, 2009 at 11:21 AM, Andreas <maps.on@gmx.net> wrote:
It's just, that windows <--> windows transfers tend to fill a 100MBit-LAN nicely with around 10MB/s. A SUSE or Knoppix on the same machine will yield only about 6-7MB/s with SAMBA and EXT3 and default settings. OK, I'm a wee bit unfair since SUSE does this on a RAID1. With EXT3 on an unRAIDed partition it pulls around 7-8.
I use and have used reiserfs for years with zero reliability problems. I regularly do 11MB/s over scp on Linux and about 10 via SAMBA. In fact, I'm going to upgrade to a gigabit switch soon and expect a whole lot faster on the way. ext3 is not a fast file system. I don't use it. I quit using ext2 once SuSE moved to Reiser for default a long time ago. ext4 would be a better alternative than ext3.
Depending on the filesystem I can watch the LAN throughput drop down in optically equal distances. It seems to transfer X MBs then pauses. There were just a couple of segments per ISO. I figure it must be at least 100MB each.
Are you using SEAGate drives? They have shown issues on lags of up to 30 seconds, which will drop an overall speed report. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wed, May 06, 2009 at 11:37:10PM -0400, Larry Stotler wrote:
Are you using SEAGate drives? They have shown issues on lags of up to 30 seconds, which will drop an overall speed report.
This might be true of consumer-grade drives, but I haven't seen or experienced in enterprise-class drives. Kurt -- Research is what I'm doing when I don't know what I'm doing. -- Wernher von Braun -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, May 7, 2009 at 12:03 AM, Kurt Wall <kwall@kurtwerks.com> wrote:
This might be true of consumer-grade drives, but I haven't seen or experienced in enterprise-class drives.
Sorry. Should have said the SEAGate SATA drives. 500-1500GB. I have 2 500's and 1 1.5 in my machine and I see occasional slow downs. A friend of mine has 2 1.5's and he's seen many more issues than I have. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009-05-07 06:03, Kurt Wall wrote:
On Wed, May 06, 2009 at 11:37:10PM -0400, Larry Stotler wrote:
Are you using SEAGate drives? They have shown issues on lags of up to 30 seconds, which will drop an overall speed report.
This might be true of consumer-grade drives, but I haven't seen or experienced in enterprise-class drives.
Actually, the firmware-bug that Seagate finally admitted to earlier this year, also affected Barracuda ES.2 drives (only SATA models though), which they classify as a enterprise product. On the consumer side, it hit several Barracuda 7200 and Maxtor DiamondMax22 models. Some reports was that the drives 'paused' on contiguous reads, others where suddenly not detected by bios, and hence 'bricked'. I have had 2 out of 3 of those drives, which where bricked due to the firmware-bug.
Kurt
/Sylvester -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, May 07, 2009 at 09:19:52AM +0200, Sylvester Lykkehus wrote:
On 2009-05-07 06:03, Kurt Wall wrote:
On Wed, May 06, 2009 at 11:37:10PM -0400, Larry Stotler wrote:
Are you using Seagate drives? They have shown issues on lags of up to 30 seconds, which will drop an overall speed report.
This might be true of consumer-grade drives, but I haven't seen or experienced in enterprise-class drives.
Actually, the firmware-bug that Seagate finally admitted to earlier this year, also affected Barracuda ES.2 drives (only SATA models though), which they classify as a enterprise product. On the consumer side, it hit several Barracuda 7200 and Maxtor DiamondMax22 models.
I stand corrected. I wasn't aware of this bug.
Some reports was that the drives 'paused' on contiguous reads, others where suddenly not detected by bios, and hence 'bricked'. I have had 2 out of 3 of those drives, which where bricked due to the firmware-bug.
Maxtor, had a nasty one a few years ago in which, under certain conditions, the head remained magnetized after completing a write and starting to move to another cylinder, resulting in corrupted data as the slightly magnetized head moved across the platter. Kurt -- Murphy's Law of Research: Enough research will tend to support your theory. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, May 7, 2009 at 3:19 AM, Sylvester Lykkehus <zly@solidonline.dk> wrote:
Actually, the firmware-bug that Seagate finally admitted to earlier this year, also affected Barracuda ES.2 drives (only SATA models though), which they classify as a enterprise product. On the consumer side, it hit several Barracuda 7200 and Maxtor DiamondMax22 models.
Didn't know about the Barracuda drives. Actually, I remember a firmware update last Noevmber(I think) that was released that bricked the drives. Thankfully I saw the issue on slashdot and didn't try that update.
Some reports was that the drives 'paused' on contiguous reads, others where suddenly not detected by bios, and hence 'bricked'. I have had 2 out of 3 of those drives, which where bricked due to the firmware-bug.
One of my 500GB drives failed on a reboot. BIOS wouldn't even see it, so I couldn't even try to flash the new firmware. I contacted Seagate and shipped it to i365. Had it back in a week and it was fixed. Thankfully, I did have a fairly complete backup this time, which I didn't need. My other 2 drives's serial numbers didn't fall into the range of problems, and I haven't had any issues with them. I do see an occasional random hanf trying to copy files between drives, but nothing really bad. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (13)
-
Andreas -
Boyd Stephen Smith Jr. -
Carlos E. R. -
Cristian Rodríguez -
Damon Register -
David C. Rankin -
Ken Schneider - openSUSE -
Kurt Wall -
Larry Stotler -
Patrick Shanahan -
Randall R Schulz -
Rikard Johnels -
Sylvester Lykkehus