Article ID: 125054, created on Mar 30, 2015, last review on Jun 17, 2016

  • Applies to:
  • Virtuozzo 6.0

Symptoms

During the backup to NFS share, the virtual machine pauses randomly:

from /var/log/parallels.log:
03-06 04:30:10.154 F /disp:5002:958638/ Vm state was changed from VMS_RUNNING to VMS_PAUSED for vm {e6a4b298-00ba-42e3-9f00-68eddebe86d6} (name='parallels.test'), to powerState 2(vpsPausedByVmFrozen)
03-06 04:30:12.935 F /disp:5002:958638/ Vm state was changed from VMS_PAUSED to VMS_RUNNING for vm {e6a4b298-00ba-42e3-9f00-68eddebe86d6} (name='parallels.test'), to powerState 0(vpsNormal)

In the meantime, Chunk Servers report outage:

from "pstorage -c $CLUSTER_NAME get-event":
06-03-15 04:30:04.340 MDS WRN: CS#1088 is inactive
06-03-15 04:30:05.340 MDS WRN: CS#1085 is inactive
06-03-15 04:30:09.022 MDS INF: Replication started, 82 chunk(s) are queued (43 chunk(s) on inactive CS#1088)
06-03-15 04:30:12.040 MDS WRN: The cluster is degraded with 26 active, 2 inactive, 0 offline CS
06-03-15 04:32:04.343 MDS INF: CS#1085 is active
06-03-15 04:32:06.343 MDS INF: CS#1088 is active
06-03-15 04:32:12.082 MDS INF: The cluster is healthy with 28 active CS
06-03-15 04:36:44.737 MDS INF: Replication completed

In the corresponding CS log the follwing stack is logged:

from /var/log/pstorage/$CLUSTER_NAME/cs-$INDEX/cs.log.gz:
06-03-15 04:30:10.760 pcs process is inactive for 9003 msecs (0)
06-03-15 04:30:10.760 [<ffffffffa046a384>] nfs_wait_bit_killable+0x24/0x40 [nfs]
06-03-15 04:30:10.760 [<ffffffffa047ab4a>] nfs_commit_inode+0xaa/0x150 [nfs]
06-03-15 04:30:10.760 [<ffffffffa046753e>] nfs_release_page+0x8e/0xb0 [nfs]
06-03-15 04:30:10.760 [<ffffffff81134150>] try_to_release_page+0x30/0x60
06-03-15 04:30:10.760 [<ffffffff81152e14>] shrink_page_list.clone.0+0x804/0xae0
06-03-15 04:30:10.760 [<ffffffff811534ab>] shrink_inactive_list+0x3bb/0xcc0
06-03-15 04:30:10.760 [<ffffffff811541e0>] shrink_lruvec+0x430/0x5f0
06-03-15 04:30:10.760 [<ffffffff81154627>] shrink_zone+0x287/0x3d0
06-03-15 04:30:10.760 [<ffffffff81154a2f>] zone_reclaim+0x2bf/0x520
06-03-15 04:30:10.760 [<ffffffff811442b4>] get_page_from_freelist+0x6d4/0x8a0
06-03-15 04:30:10.760 [<ffffffff81146762>] __alloc_pages_nodemask+0x122/0xb20
06-03-15 04:30:10.760 [<ffffffff8118c04a>] alloc_pages_current+0xaa/0x110
06-03-15 04:30:10.760 [<ffffffff814b3d69>] tcp_sendmsg+0xa59/0xb20
06-03-15 04:30:10.760 [<ffffffff81455a17>] sock_sendmsg+0x117/0x140
06-03-15 04:30:10.760 [<ffffffff81456249>] sys_sendto+0x139/0x190
06-03-15 04:30:10.760 [<ffffffff8100b102>] system_call_fastpath+0x16/0x1b
06-03-15 04:30:10.760 [<ffffffffffffffff>] 0xffffffffffffffff

Cause

Virtual machine backup to the slow NFS share may lead to CS inactive/active switching. The issue is recognized as a product bug with internal IDs PSBM-28058 and PSBM-27322.

Resolution

PSBM-28058 was fixed in PCS6.0 Update8 Hotfix2.

The fix for PSBM-27322 will be included in the future updates of PCS.

Related articles

119953- Virtual machine cannot be resumed - problem reading the virtual machine location

123228- Virtual Machines switch between PAUSED and RUNNING status

Search Words

backups

is inactive

Vm state was changed from VMS_RUNNING to VMS_PAUSED

virtual machne pauses

backup

nfs

c62e8726973f80975db0531f1ed5c6a2 2897d76d56d2010f4e3a28f864d69223 0dd5b9380c7d4884d77587f3eb0fa8ef

Email subscription for changes to this article
Save as PDF