I have check before and after OOM-killer slabinfo There have some object have high counts. And dm-snapshot-ex always have high counts. Have any explain about those value? Kevin Cheng ---------------------------------------------------------------------------- ------------------------------------------------- Before: kcopyd-jobs 512 525 264 15 1 : tunables 54 27 0 : slabdata 35 35 0 After : kcopyd-jobs 16524 62685 264 15 1 : tunables 54 27 0 : slabdata 4179 4179 0 Before: dm-snapshot-in 128 177 64 59 1 : tunables 120 60 0 : slabdata 3 3 0 After : dm-snapshot-in 24833 73219 64 59 1 : tunables 120 60 0 : slabdata 1241 1241 0 Before: dm-snapshot-ex 780545 780680 24 145 1 : tunables 120 60 0 : slabdata 5384 5384 0 Before: dm_tio 5124 5278 16 203 1 : tunables 120 60 0 : slabdata 26 26 0 Before: dm_io 5133 5239 20 169 1 : tunables 120 60 0 : slabdata 31 31 0 After : dm_tio 23337 32277 16 203 1 : tunables 120 60 0 : slabdata 159 159 0 After : dm_io 23339 32448 20 169 1 : tunables 120 60 0 : slabdata 192 192 0 Before: journal_head 2 72 52 72 1 : tunables 120 60 0 : slabdata 1 1 0 After : journal_head 23029 25272 52 72 1 : tunables 120 60 0 : slabdata 351 351 0 Before: biovec-4 39 59 64 59 1 : tunables 120 60 0 : slabdata 1 1 0 Before: biovec-1 76 406 16 203 1 : tunables 120 60 0 : slabdata 2 2 0 Before: bio 330 330 128 30 1 : tunables 120 60 0 : slabdata 11 11 0 After : biovec-4 2815 3717 64 59 1 : tunables 120 60 0 : slabdata 63 63 0 After : biovec-1 33507 47096 16 203 1 : tunables 120 60 0 : slabdata 232 232 0 After : bio 36390 50130 128 30 1 : tunables 120 60 0 : slabdata 1671 1671 0 Before: buffer_head 16360 16416 52 72 1 : tunables 120 60 0 : slabdata 228 228 0 After : buffer_head 24234 36000 52 72 1 : tunables 120 60 0 : slabdata 500 500 0 -----Original Message----- From: Andrew Morton [mailto:akpm@xxxxxxxx] Sent: Thursday, September 14, 2006 1:37 PM To: kevin Cheng Cc: 'bugme-daemon@xxxxxxxxxxxxxxxxxxxx'; dm-devel@xxxxxxxxxx; agk@xxxxxxxxxx Subject: Re: [Bugme-new] [Bug 7158] New: Out of memory happen in snapshot On Thu, 14 Sep 2006 13:25:52 +0800 "kevin Cheng" <kevin_cheng@xxxxxxxxxx> wrote: > Dear Andrew: > Thanks for your message and help. > > Attech file is the slabinfo from my machine. Thanks!! > OK, thanks. A number of DM-related slab caches have really high object counts. So there's a good chance that this is either a mistuning/imbalance issue in DM or an outright leak. I'd ask the DM developers to take over here please. -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
