When a file is unlinked before it was accounted by the marker, then we may see error "ctx for the node ... is NULL" at many places. It is better to set the inode ctx before lookup is unwind back to protocol server.
Patch submitted upstream: http://review.gluster.org/#/c/13748/
This issue was seen at Goldman Sachs setups and we are providing a hotfix. Patch is already merged upstream: http://review.gluster.org/#/c/13748/. It is good to include the fix in 3.1.3 as well. Thanks, Vijay
This bug was accidentally moved from POST to MODIFIED via an error in automation, please see mmccune with any questions
*** Bug 1320925 has been marked as a duplicate of this bug. ***
*** Bug 1320926 has been marked as a duplicate of this bug. ***
Patches merged: https://code.engineering.redhat.com/gerrit/71535 https://code.engineering.redhat.com/gerrit/71537 https://code.engineering.redhat.com/gerrit/71539 https://code.engineering.redhat.com/gerrit/71540
Set network.inode-lru-limit 1 Enabled quota and limit usage Created some file on client and deleted it. Not seeing any errors in bricks logs. Bug verified on build glusterfs-3.7.9-2.el7rhgs.x86_64.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2016:1240