Bug 1332845

Summary:	Disperse volume fails on high load and logs show some assertion failures
Product:	[Community] GlusterFS	Reporter:	Xavi Hernandez <jahernan>
Component:	disperse	Assignee:	Xavi Hernandez <jahernan>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	3.8.0	CC:	aspandey, bugs, pkarampu
Target Milestone:	---	Keywords:	Reopened, Triaged
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	glusterfs-3.8rc2	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:	1331254	Environment:
Last Closed:	2016-06-16 14:05:07 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	1331254, 1339465
Bug Blocks:

Description Xavi Hernandez 2016-05-04 08:10:13 UTC

+++ This bug was initially created as a clone of Bug #1331254 +++

+++ This bug was initially created as a clone of Bug #1330132 +++

Description of problem:

A distributed iozone test over multiple NFS mounts on different machines causes the test to fail and some assertion failures appear on the logs:

[2016-04-21 19:29:58.096645] E [ec-inode-read.c:1157:ec_readv_rebuild] (-->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(__ec_manager+0x5b) [0x7f9e4e8f18bb] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_manager_readv+0x107) [0x7f9e4e908197] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_readv_rebuild+0x236) [0x7f9e4e907f26] ) 0-: Assertion failed: ec_get_inode_size(fop, fop->fd->inode, &cbk->iatt[0].ia_size)
[2016-04-21 19:29:58.126547] E [ec-common.c:1641:ec_lock_unfreeze] (-->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_manager_inodelk+0x155) [0x7f9e4e8fc305] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_unlocked+0x35) [0x7f9e4e8f3c25] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_lock_unfreeze+0x100) [0x7f9e4e8f3ab0] ) 0-: Assertion failed: list_empty(&lock->waiting) && list_empty(&lock->owners)
[2016-04-21 19:30:05.998568] E [ec-inode-read.c:1612:ec_manager_stat] (-->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_resume+0x88) [0x7f9e4e8f1a68] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(__ec_manager+0x5b) [0x7f9e4e8f18bb] -->/usr/lib64/glusterfs/3.7.10/xlator/cluster/disperse.so(ec_manager_stat+0x315) [0x7f9e4e905ed5] ) 0-: Assertion failed: ec_get_inode_size(fop, fop->locks[0].lock->loc.inode, &cbk->iatt[0].ia_size)
[2016-04-21 19:30:05.999146] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-8: remote operation failed [Invalid argument]
[2016-04-21 19:30:05.999132] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-10: remote operation failed [Invalid argument]
[2016-04-21 19:30:05.999237] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-11: remote operation failed [Invalid argument]
[2016-04-21 19:30:05.999259] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-7: remote operation failed [Invalid argument]
[2016-04-21 19:30:05.999326] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-9: remote operation failed [Invalid argument]
[2016-04-21 19:30:06.047496] E [MSGID: 114031] [client-rpc-fops.c:1624:client3_3_inodelk_cbk] 0-test-client-6: remote operation failed [Invalid argument]
[2016-04-21 19:30:06.047559] W [MSGID: 122015] [ec-common.c:1675:ec_unlocked] 0-test-disperse-1: entry/inode unlocking failed (FSTAT) [Invalid argument]

Version-Release number of selected component (if applicable): mainline


How reproducible:

It happens randomly after some time running the distributed iozone test.

Steps to Reproduce:
1.
2.
3.

Actual results:

Volume access fails and iozone quits with an error.

Expected results:

iozone should complete the test successfully.

Additional info:

Probably related to a race when cancelling the lock release timeout while the callback is already executing. In this case the new fop is not placed in the right waiting list.

--- Additional comment from Vijay Bellur on 2016-04-29 11:19:27 CEST ---

REVIEW: http://review.gluster.org/14112 (cluster/ec: Fix issues with eager locking) posted (#1) for review on master by Xavier Hernandez (xhernandez)

--- Additional comment from Vijay Bellur on 2016-05-02 16:45:05 CEST ---

COMMIT: http://review.gluster.org/14112 committed in master by Jeff Darcy (jdarcy) 
------
commit 209985e861f4d8a22bfdb457c0e8d7045ab44553
Author: Xavier Hernandez <xhernandez>
Date:   Thu Apr 28 08:42:40 2016 +0200

    cluster/ec: Fix issues with eager locking
    
    Due to a race in timer cancellation, in some cases it was possible
    to unlock the lock while another concurrent fop that needed it
    continues execution as if it were not released.
    
    This patch also fixes an issue that caused a lock to not be released
    if an error was found while preparing ec_update_size_version().
    
    Change-Id: I1344a3f5ecfc333f05a09e62653838264c9c26b1
    BUG: 1331254
    Signed-off-by: Xavier Hernandez <xhernandez>
    Reviewed-on: http://review.gluster.org/14112
    Smoke: Gluster Build System <jenkins.com>
    CentOS-regression: Gluster Build System <jenkins.com>
    Reviewed-by: Chen Chen <chenchen>
    NetBSD-regression: NetBSD Build System <jenkins.org>

Comment 1 Vijay Bellur 2016-05-04 08:12:08 UTC

REVIEW: http://review.gluster.org/14206 (cluster/ec: Fix issues with eager locking) posted (#1) for review on release-3.8 by Xavier Hernandez (xhernandez)

Comment 2 Vijay Bellur 2016-05-06 14:31:47 UTC

COMMIT: http://review.gluster.org/14206 committed in release-3.8 by Niels de Vos (ndevos) 
------
commit d1e0200f7dbbc412b8fc0127be2324beaade1c78
Author: Xavier Hernandez <xhernandez>
Date:   Thu Apr 28 08:42:40 2016 +0200

    cluster/ec: Fix issues with eager locking
    
    Due to a race in timer cancellation, in some cases it was possible
    to unlock the lock while another concurrent fop that needed it
    continues execution as if it were not released.
    
    This patch also fixes an issue that caused a lock to not be released
    if an error was found while preparing ec_update_size_version().
    
    > Change-Id: I1344a3f5ecfc333f05a09e62653838264c9c26b1
    > BUG: 1331254
    > Signed-off-by: Xavier Hernandez <xhernandez>
    > Reviewed-on: http://review.gluster.org/14112
    > Smoke: Gluster Build System <jenkins.com>
    > CentOS-regression: Gluster Build System <jenkins.com>
    > Reviewed-by: Chen Chen <chenchen>
    > NetBSD-regression: NetBSD Build System <jenkins.org>
    
    Change-Id: I9ccd585a9b9952b6787cfca6720bc59b9c8ddab9
    BUG: 1332845
    Signed-off-by: Xavier Hernandez <xhernandez>
    Reviewed-on: http://review.gluster.org/14206
    Smoke: Gluster Build System <jenkins.com>
    NetBSD-regression: NetBSD Build System <jenkins.org>
    CentOS-regression: Gluster Build System <jenkins.com>
    Reviewed-by: Niels de Vos <ndevos>

Comment 3 Niels de Vos 2016-06-16 14:05:07 UTC

This bug is getting closed because a release has been made available that should address the reported issue. In case the problem is still not fixed with glusterfs-3.8.0, please open a new bug report.

glusterfs-3.8.0 has been announced on the Gluster mailinglists [1], packages for several distributions should become available in the near future. Keep an eye on the Gluster Users mailinglist [2] and the update infrastructure for your distribution.

[1] http://blog.gluster.org/2016/06/glusterfs-3-8-released/
[2] http://thread.gmane.org/gmane.comp.file-systems.gluster.user