RHBA-2018:0083 - Bug Fix Advisory

Topic

Updated glusterfs packages that fix several bugs are now available for Red Hat Gluster Storage 3.3 Update 1 on Red Hat Enterprise Linux 7.

Description

Red Hat Gluster Storage is a software only scale-out storage solution that
provides flexible and affordable unstructured data storage. It unifies data
storage and infrastructure, increases performance, and improves
availability and manageability to meet enterprise-level storage challenges.

This advisory addresses critical bugs related to memory leaks and fixes
the following bugs:

• Executing gluster volume set operations on volumes from gluster CLI

caused a significant memory leak in glusterd process. The memory leak
issues have been fixed in this release. (BZ#1526363)

• Previously, while executing a series of gluster commands like volume

create/start/stop/delete, glusterd process' memory footprint increased.
In a scalable environment where multiple volumes are created, deleted,
and recreated led to high glusterd memory consumption. With this update,
the identified memory leaks in glusterd are fixed. (BZ#1526365, BZ#1526374)

• Executing multiple volume commands concurrently on the same volume,

from different peers of the trusted storage pool, results in one of
the glusterd processes from the same pool moving to the locked state.
Any volume management operations performed on the same volume fail until
the glusterd service is restarted on the node where glusterd is in
transaction locked state. This release introduces a default transaction
lock timeout of 3 minutes. Any glusterd process reaching the locked state
will remain there for 3 minutes. Any transactions made now execute
successfully after this timeout period. (BZ#1526372)

• The portmap entry allocation that glusterd maintains for each of its

brick is cleaned up on a graceful shutdown of the respective brick
processes. When a brick process was killed through SIGKILL signal or
crashed, glusterd did not clean up the respective portmap entry allocation
which could result in two portmap allocations for the same brick when the
brick process was restarted. glusterd reports with a stale port to the
client, resulting failure in connection. With this fix, the stale port
entry is removed even for a brick crash or SIGKILL event of a brick
process. The client now successfully connects to the brick after the brick
is restarted. (BZ#1526371)

• When processes that use a large number of POSIX locks, like Samba,

were used in combination with the gluster clear-locks command, a memory
leak created high memory consumption on brick processes. This sometimes
triggered OOM killer on brick processes. The lock translator has been
updated to correct this issue and fix the memory leak. (BZ#1526377)

Users of glusterfs are advised to upgrade to these updated packages, which fix these bugs.

Solution

Before applying this update, make sure all previously released errata
relevant to your system have been applied.

For details on how to apply this update, refer to:

https://access.redhat.com/articles/11258

Affected Products

• Red Hat Enterprise Linux Server 7 x86_64
• Red Hat Virtualization 4 for RHEL 7 x86_64
• Red Hat Gluster Storage Server for On-premise 3 for RHEL 7 x86_64

Fixes

• BZ - 1526363 - [RHEL 7] high memory usage by glusterd on executing gluster volume set operations
• BZ - 1526368 - Brick Multiplexing: stale brick processes getting created and volume status shows brick as down(pkill glusterfsd glusterfs ,glusterd restart)
• BZ - 1526371 - clean up port map on brick disconnect
• BZ - 1526372 - Unable to acquire lock for gluster volume leading to 'another transaction in progress' error
• BZ - 1526373 - Brick Multiplexing: Gluster volume start force complains with command "Error : Request timed out" when there are multiple volumes
• BZ - 1527147 - [GSS] [Regression] glusterfs threads consuming swap space until it runs out
• BZ - 1527772 - remove ExclusiveArch directive from SPEC file
• BZ - 1530217 - Brick multiplexing: glustershd fails to start on a volume force start after a brick is down
• BZ - 1530320 - Brick Multiplexing: brick still down in heal info context(glfs) even though brick is online

CVEs

(none)

References

(none)