atom feed82 messages in org.freebsd.freebsd-fsZFS patches.
FromSent OnAttachments
Pawel Jakub DawidekJul 27, 2008 12:53 pm 
Arne SchwabeJul 27, 2008 2:05 pm 
Fabian KeilJul 27, 2008 2:20 pm 
Wes MorganJul 27, 2008 2:48 pm 
Ulf LilleengenJul 27, 2008 3:27 pm 
Ulf LilleengenJul 27, 2008 3:31 pm 
Norikatsu ShigemuraJul 27, 2008 6:11 pm 
Xin LIJul 27, 2008 6:32 pm 
Xin LIJul 27, 2008 6:37 pm 
Max LaierJul 27, 2008 6:46 pm 
Ivan VorasJul 27, 2008 9:24 pm 
Ollivier RobertJul 27, 2008 9:46 pm 
Wes MorganJul 27, 2008 10:56 pm 
Norikatsu ShigemuraJul 28, 2008 3:02 am 
David GrochowskiJul 28, 2008 5:21 am 
Pawel Jakub DawidekJul 28, 2008 8:32 am 
Max LaierJul 28, 2008 12:54 pm 
Pawel Jakub DawidekJul 28, 2008 12:57 pm 
OutBackdingoJul 28, 2008 1:26 pm 
John NielsenJul 28, 2008 4:15 pm 
Sam Fourman Jr.Jul 28, 2008 4:19 pm 
Philip M. GollucciJul 28, 2008 7:49 pm 
Beat GätziJul 28, 2008 8:05 pm 
Jorn ArgeloJul 28, 2008 8:22 pm 
Pawel Jakub DawidekJul 28, 2008 8:53 pm 
Sven WJul 28, 2008 8:55 pm 
Max LaierJul 28, 2008 9:16 pm 
Antony MawerJul 28, 2008 9:28 pm 
Pawel Jakub DawidekJul 28, 2008 9:34 pm 
Ivan VorasJul 29, 2008 12:14 am 
Antony MawerJul 29, 2008 12:41 am 
Ivan VorasJul 29, 2008 12:50 am 
Remy NonnenmacherJul 29, 2008 1:33 am 
Adam McDougallJul 29, 2008 1:52 am 
Gary JennejohnJul 29, 2008 3:56 am 
Gary JennejohnJul 29, 2008 4:58 am 
Stefan LambrevJul 29, 2008 5:26 am 
Jeremy ChadwickJul 29, 2008 5:55 am 
OutBackdingoJul 29, 2008 6:40 am 
Randy BushJul 29, 2008 6:51 am 
OutBackdingoJul 29, 2008 9:10 am 
George HartzellJul 29, 2008 9:24 am 
George HartzellJul 29, 2008 9:36 am 
George HartzellJul 29, 2008 10:25 am 
Zaphod BeeblebroxJul 29, 2008 12:22 pm 
Jeremy ChadwickJul 29, 2008 1:17 pm 
Ulf LilleengenJul 29, 2008 2:11 pm 
Ulf LilleengenJul 29, 2008 2:16 pm 
Aristedes ManiatisJul 30, 2008 2:28 am 
CZUCZY GergelyJul 30, 2008 2:44 am 
Pawel Jakub DawidekJul 30, 2008 2:53 am 
Norikatsu ShigemuraJul 30, 2008 9:32 am 
Pascal HofsteeJul 30, 2008 11:40 am 
Claus GuttesenJul 31, 2008 8:07 am 
kevinJul 31, 2008 9:03 am 
Benjamin CloseJul 31, 2008 4:24 pm 
Claus GuttesenAug 1, 2008 5:19 am 
Claus GuttesenAug 1, 2008 6:38 am 
George HartzellAug 3, 2008 1:59 pm 
Norikatsu ShigemuraAug 5, 2008 8:45 am 
John BaldwinAug 5, 2008 10:27 am 
Claus GuttesenAug 6, 2008 3:14 am 
Norikatsu ShigemuraAug 6, 2008 10:35 am 
John NielsenAug 13, 2008 10:48 am 
Larry RosenmanAug 17, 2008 12:25 pm 
Larry RosenmanAug 17, 2008 3:13 pm 
Pascal HofsteeAug 20, 2008 9:19 am 
Weldon S Godfrey 3Aug 20, 2008 11:43 am 
Larry RosenmanAug 20, 2008 12:10 pm 
Weldon S Godfrey 3Aug 20, 2008 12:22 pm 
Lorenzo PeroneAug 21, 2008 5:43 am 
Matt SimersonAug 21, 2008 11:14 am 
Oliver FrommeAug 21, 2008 1:16 pm 
Dag-Erling SmørgravAug 23, 2008 4:41 am 
Stefan LambrevAug 25, 2008 4:23 am 
swel...@gmail.comAug 28, 2008 4:29 pm.rej, .rej, .rej
Pawel Jakub DawidekAug 29, 2008 12:47 am 
Larry RosenmanAug 30, 2008 1:34 pm 
Peter JeremyOct 31, 2008 1:17 pm 
Marcus MüllerNov 8, 2008 8:50 am 
Larry RosenmanNov 8, 2008 1:09 pm 
Dimitar VasilevNov 8, 2008 8:36 pm 
Subject:ZFS patches.
From:Ivan Voras (ivo@freebsd.org)
Date:Jul 29, 2008 12:50:53 am
List:org.freebsd.freebsd-fs

Antony Mawer wrote:

Ivan Voras wrote:

I currently don't have high-end (4 CPU+) AMD64 machines to test, but with 1 CPU i386 virtual machine in VMWare, with 1 GB of memory, kmem_size=kmem_size_max=512M and no other tuning, with latest zpool format (11) it took about 15 minutes to get a "kmem_map too small" panic on a mixed load (buildkernel + blogbench + bonnie++).

I've then tried the same load on the "real" hardware, 2 CPU, 2 GB memory, kmem_size=kmem_size_max=512M, and no other tuning, with the older zpool format (6) i get the same panic, though it takes about twice as long to happen.

Have you tried tuning arc_max and/or monitoring vmstat -m to see what is happening? What does arc_max get auto-tuned to at the moment (ie. without manually specifying)?

One of the things I recall reading that arc_max is more like a guide, as some ZFS threads can exceed the max whilst other thread(s) go around cleaning up and freeing memory once the limit is hit.

Maybe some better smarts are needed in auto-tuning arc_max so that it leaves more of a buffer zone than it does at the moment...?

I think speculation in the same direction was discussed with the original port of ZFS - I don't know the details but if it arc_max could be better auto-tuned, I think it should be by now.

I'm more concerned about the "quiet period" before the panic. I notice ZFS threads have the same priority (PRI field in top) as userland threads (e.g. 44, 55...), while GEOM threads have it different (-8). I don't have the McCusicks book about it so I don't know how priorities whould work, but is this situation OK?

I will monitor vmstate -m if it will help Pawel. Should I?