Bug 1779284

Summary: Backport GNFS memory leak fix to version 5
Product: [Community] GlusterFS Reporter: Andreas Loibl <mail>
Component: nfsAssignee: bugs <bugs>
Status: CLOSED NEXTRELEASE QA Contact:
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 5CC: atumball, bugs, rhs-bugs, sankarshan, skoduri, vbellur, xiechanglong
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: ---
Doc Text:
Story Points: ---
Clone Of: 1711240 Environment:
Last Closed: 2019-12-06 10:32:56 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
memory usage graph, glusterfs 5.10 with and without the patch applied none

Description Andreas Loibl 2019-12-03 16:44:39 UTC 
I'm opening this issue to track a backport of the
  Change-Id: Ic61a8fdd0e941e136c98376a87b5a77fa8c22316
back into release-5 branch.

The bug #1711240 has been fixed and a patch committed and merged to master:

https://bugzilla.redhat.com/show_bug.cgi?id=1711240

https://review.gluster.org/22738 (inode: fix wrong loop count in __inode_ctx_free)


Reason for backporting the change:

for the master branch where this fix was merged into (i.e. version 7) GNFS is deprecated and there is "--disable-gnfs" as default for (at least) the debian packaging now: https://github.com/gluster/glusterfs-debian/commit/c91ec77552

so this fix for the GNFS memory leak is not actually accessible by upgrading to the latest version, it will be neccessary to manually rebuild it (either apply the patch for the previous releases or set "--enable-gnfs" configure flags for version 7)

if this change gets backported everyone could still use the official glusterfs packages from the repository instead of having to maintain own, patched versions.

the patch applies cleanly on version 5 and resolves the issue, I've tested it using glusterfs 5.10, manually rebuilt with the patch applied, and no more leaking of memory occurred.
Comment 1 Worker Ant 2019-12-03 16:48:01 UTC 
REVIEW: https://review.gluster.org/23802 (inode: fix wrong loop count in __inode_ctx_free) posted (#1) for review on release-5 by None
Comment 2 Andreas Loibl 2019-12-03 17:27:21 UTC 
Created attachment 1641740 [details]
memory usage graph, glusterfs 5.10 with and without the patch applied
Comment 3 Worker Ant 2019-12-06 10:32:56 UTC 
REVIEW: https://review.gluster.org/23802 (inode: fix wrong loop count in __inode_ctx_free) merged (#2) on release-5 by hari gowtham