Hi Wido,
Thanks for the information and let us know if this is a bug.
As workaround we will go with small bluestore_cache_size to 100MB.
Thanks,
Muthu
On 16 February 2017 at 14:04, Wido den Hollander <wido@xxxxxxxx> wrote:
> Op 16 februari 2017 om 7:19 schreef Muthusamy Muthiah <muthiah.muthusamy@xxxxxxxxx>:
>
>
> Thanks IIya Letkowski for the information we will change this value
> accordingly.
>
What I understand from yesterday's performance meeting is that this seems like a bug. Lowering this buffer reduces memory, but the root-cause seems to be memory not being freed. A few bytes of a larger allocation still allocated causing this buffer not to be freed.
Tried:
debug_mempools = true
$ ceph daemon osd.X dump_mempools
Might want to view the YouTube video of yesterday when it's online: https://www.youtube.com/channel/UCno- Fry25FJ7B4RycCxOtfw/videos
Wido
> Thanks,
> Muthu
>
> On 15 February 2017 at 17:03, Ilya Letkowski <mj12.svetzari@xxxxxxxxx>
> wrote:
>
> > Hi, Muthusamy Muthiah
> >
> > I'm not totally sure that this is a memory leak.
> > We had same problems with bluestore on ceph v11.2.0.
> > Reduce bluestore cache helped us to solve it and stabilize OSD memory
> > consumption on the 3GB level.
> >
> > Perhaps this will help you:
> >
> > bluestore_cache_size = 104857600
> >
> >
> >
> > On Tue, Feb 14, 2017 at 11:52 AM, Muthusamy Muthiah <
> > muthiah.muthusamy@xxxxxxxxx> wrote:
> >
> >> Hi All,
> >>
> >> On all our 5 node cluster with ceph 11.2.0 we encounter memory leak
> >> issues.
> >>
> >> Cluster details : 5 node with 24/68 disk per node , EC : 4+1 , RHEL 7.2
> >>
> >> Some traces using sar are below and attached the memory utilisation graph
> >> .
> >>
> >> (16:54:42)[cn2.c1 sa] # sar -r
> >> 07:50:01 kbmemfree kbmemused %memused kbbuffers kbcached kbcommit %commit
> >> kbactive kbinact kbdirty
> >> 10:20:01 32077264 132754368 80.54 16176 3040244 77767024 47.18 51991692
> >> 2676468 260
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> *10:30:01 32208384 132623248 80.46 16176 3048536 77832312 47.22 51851512
> >> 2684552 1210:40:01 32067244 132764388 80.55 16176 3059076 77832316 47.22
> >> 51983332 2694708 26410:50:01 30626144 134205488 81.42 16176 3064340
> >> 78177232 47.43 53414144 2693712 411:00:01 28927656 135903976 82.45 16176
> >> 3074064 78958568 47.90 55114284 2702892 1211:10:01 27158548 137673084 83.52
> >> 16176 3080600 80553936 48.87 56873664 2708904 1211:20:01 26455556 138376076
> >> 83.95 16176 3080436 81991036 49.74 57570280 2708500 811:30:01 26002252
> >> 138829380 84.22 16176 3090556 82223840 49.88 58015048 2718036 1611:40:01
> >> 25965924 138865708 84.25 16176 3089708 83734584 50.80 58049980 2716740
> >> 1211:50:01 26142888 138688744 84.14 16176 3089544 83800100 50.84 57869628
> >> 2715400 16*
> >>
> >> ...
> >> ...
> >>
> >> In the attached graph, there is increase in memory utilisation by
> >> ceph-osd during soak test. And when it reaches the system limit of 128GB
> >> RAM , we could able to see the below dmesg logs related to memory out when
> >> the system reaches close to 128GB RAM. OSD.3 killed due to Out of memory
> >> and started again.
> >>
> >> [Tue Feb 14 03:51:02 2017] *tp_osd_tp invoked oom-killer:
> >> gfp_mask=0x280da, order=0, oom_score_adj=0*
> >> [Tue Feb 14 03:51:02 2017] tp_osd_tp cpuset=/ mems_allowed=0-1
> >> [Tue Feb 14 03:51:02 2017] CPU: 20 PID: 11864 Comm: tp_osd_tp Not tainted
> >> 3.10.0-327.13.1.el7.x86_64 #1
> >> [Tue Feb 14 03:51:02 2017] Hardware name: HP ProLiant XL420 Gen9/ProLiant
> >> XL420 Gen9, BIOS U19 09/12/2016
> >> [Tue Feb 14 03:51:02 2017] ffff8819ccd7a280 0000000030e84036
> >> ffff881fa58f7528 ffffffff816356f4
> >> [Tue Feb 14 03:51:02 2017] ffff881fa58f75b8 ffffffff8163068f
> >> ffff881fa3478360 ffff881fa3478378
> >> [Tue Feb 14 03:51:02 2017] ffff881fa58f75e8 ffff8819ccd7a280
> >> 0000000000000001 000000000001f65f
> >> [Tue Feb 14 03:51:02 2017] Call Trace:
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff816356f4>] dump_stack+0x19/0x1b
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8163068f>] dump_header+0x8e/0x214
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8116ce7e>]
> >> oom_kill_process+0x24e/0x3b0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8116c9e6>] ?
> >> find_lock_task_mm+0x56/0xc0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8116d6a6>]
> >> *out_of_memory+0x4b6/0x4f0*
> >> Feb 14 03:51:40 fr-paris kernel: *Out of memory: Kill process 7657> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81173885>]
> >> __alloc_pages_nodemask+0xa95/0xb90
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811b792a>]
> >> alloc_pages_vma+0x9a/0x140
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811976c5>]
> >> handle_mm_fault+0xb85/0xf50
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811957fb>] ?
> >> follow_page_mask+0xbb/0x5c0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81197c2b>]
> >> __get_user_pages+0x19b/0x640
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8119843d>]
> >> get_user_pages_unlocked+0x15d/0x1f0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8106544f>]
> >> get_user_pages_fast+0x9f/0x1a0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8121de78>]
> >> do_blockdev_direct_IO+0x1a78/0x2610
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81218c40>] ? I_BDEV+0x10/0x10
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8121ea65>]
> >> __blockdev_direct_IO+0x55/0x60
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81218c40>] ? I_BDEV+0x10/0x10
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81219297>]
> >> blkdev_direct_IO+0x57/0x60
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81218c40>] ? I_BDEV+0x10/0x10
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8116af63>]
> >> generic_file_aio_read+0x6d3/0x750
> >> [Tue Feb 14 03:51:02 2017] [<ffffffffa038ad5c>] ?
> >> xfs_iunlock+0x11c/0x130 [xfs]
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811690db>] ? unlock_page+0x2b/0x30
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81192f21>] ? __do_fault+0x401/0x510
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff8121970c>] blkdev_aio_read+0x4c/0x70
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811ddcfd>] do_sync_read+0x8d/0xd0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811de45c>] vfs_read+0x9c/0x170
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff811df182>] SyS_pread64+0x92/0xc0
> >> [Tue Feb 14 03:51:02 2017] [<ffffffff81645e89>]
> >> system_call_fastpath+0x16/0x1b
> >>
> >>
> >> (ceph-osd) score 45 or sacrifice child*
> >> Feb 14 03:51:40 fr-paris kernel: Killed process 7657 (ceph-osd)
> >> total-vm:8650208kB, anon-rss:6124660kB, file-rss:1560kB
> >> Feb 14 03:51:41 fr-paris systemd:* ceph-osd@3.service: main process
> >> exited, code=killed, status=9/KILL*
> >> Feb 14 03:51:41 fr-paris systemd: Unit ceph-osd@3.service entered failed
> >> state.
> >> Feb 14 03:51:41 fr-paris systemd: *ceph-osd@3.service failed.*
> >> Feb 14 03:51:41 fr-paris systemd: cassandra.service: main process exited,
> >> code=killed, status=9/KILL
> >> Feb 14 03:51:41 fr-paris systemd: Unit cassandra.service entered failed
> >> state.
> >> Feb 14 03:51:41 fr-paris systemd: cassandra.service failed.
> >> Feb 14 03:51:41 fr-paris ceph-mgr: 2017-02-14 03:51:41.978878
> >> 7f51a3154700 -1 mgr ms_dispatch osd_map(7517..7517 src has 6951..7517) v3
> >> Feb 14 03:51:42 fr-paris systemd: Device dev-disk-by\x2dpartlabel-ceph\x5cx20block.device > >> Feb 14 03:52:02 fr-paris systemd: *Starting Ceph object storage daemon
> >> appeared twice with different sysfs paths /sys/devices/pci0000:00/0000:0
> >> 0:03.2/0000:03:00.0/host0/target0:0:0/0:0:0:9/block/sdj/ sdj2 and
> >> /sys/devices/pci0000:00/0000:00:03.2/0000:03:00.0/host0/ targ
> >> et0:0:0/0:0:0:4/block/sde/sde2
> >> Feb 14 03:51:42 fr-paris ceph-mgr: 2017-02-14 03:51:42.992477
> >> 7f51a3154700 -1 mgr ms_dispatch osd_map(7518..7518 src has 6951..7518) v3
> >> Feb 14 03:51:43 fr-paris ceph-mgr: 2017-02-14 03:51:43.508990
> >> 7f51a3154700 -1 mgr ms_dispatch mgrdigest v1
> >> Feb 14 03:51:48 fr-paris ceph-mgr: 2017-02-14 03:51:48.508970
> >> 7f51a3154700 -1 mgr ms_dispatch mgrdigest v1
> >> Feb 14 03:51:53 fr-paris ceph-mgr: 2017-02-14 03:51:53.509592
> >> 7f51a3154700 -1 mgr ms_dispatch mgrdigest v1
> >> Feb 14 03:51:58 fr-paris ceph-mgr: 2017-02-14 03:51:58.509936
> >> 7f51a3154700 -1 mgr ms_dispatch mgrdigest v1
> >> Feb 14 03:52:01 fr-paris systemd: ceph-osd@3.service holdoff time over,
> >> scheduling restart.
> >> osd.3.*..
> >> Feb 14 03:52:02 fr-paris systemd: Started Ceph object storage daemon
> >> osd.3.
> >> Feb 14 03:52:02 fr-paris numactl: 2017-02-14 03:52:02.307106 7f1e499bb940
> >> -1 WARNING: the following dangerous and experimental features are enabled:
> >> bluestore,rocksdb
> >> Feb 14 03:52:02 fr-paris numactl: 2017-02-14 03:52:02.317687 7f1e499bb940
> >> -1 WARNING: the following dangerous and experimental features are enabled:
> >> bluestore,rocksdb
> >> Feb 14 03:52:02 fr-paris numactl: starting osd.3 at - osd_data
> >> /var/lib/ceph/osd/ceph-3 /var/lib/ceph/osd/ceph-3/journal
> >> Feb 14 03:52:02 fr-paris numactl: 2017-02-14 03:52:02.333522 7f1e499bb940
> >> -1 WARNING: experimental feature 'bluestore' is enabled
> >> Feb 14 03:52:02 fr-paris numactl: Please be aware that this feature is
> >> experimental, untested,
> >> Feb 14 03:52:02 fr-paris numactl: unsupported, and may result in data
> >> corruption, data loss,
> >> Feb 14 03:52:02 fr-paris numactl: and/or irreparable damage to your
> >> cluster. Do not use
> >> Feb 14 03:52:02 fr-paris numactl: feature with important data.
> >>
> >> This seems to happen only in 11.2.0 and not in 11.1.x . Could you please
> >> help us in resolving this issue by means of any config change to limit the
> >> memory use on ceph-osd or a bug in the current kraken release.
> >>
> >> Thanks,
> >> Muthu
> >>
> >> _______________________________________________
> >> ceph-users mailing list
> >> ceph-users@xxxxxxxxxxxxxx
> >> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph. com
> >>
> >>
> >
> >
> > --
> > С уважением / Best regards
> >
> > Илья Летковский / Ilya Letkouski
> >
> > Phone, Viber: +375 29 3237335
> >
> > Minsk, Belarus (GMT+3)
> >
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph. com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
