RHBA-2020:0621 - Bug Fix Advisory

Topic

Updated OpenShift Container Storage packages that fix several bugs are now available for Red Hat OpenShift Container Storage 3.11 Update 5.

Description

Red Hat OpenShift Container Storage (OCS) is a provider of agnostic persistent storage for OpenShift Container Platform either in-house or in a
hybrid cloud. As a Red Hat storage solution, OCS is completely integrated
with OpenShift Container Platform for deployment, management, and
monitoring.

This advisory fixes the following bugs:

• When the block volumes are deleted, file or files are left under `/<brick-dir>/.glusterfs/unlink/' of the brick, wasting space. For every volume deleted, one file of 64MB is kept on every brick. Assuming there are 100 Block volumes created on a given BHV and deleted, the 64MB x 100 = 6.4GB space will be wasted on every brick. With this update, tcmu-runner is restarted and the open fd's in tcmu-runner daemon is closed and the files under ‘/brick/.glusterfs/unlink/' are removed, freeing the occupied space. The block volume creations are done successfully. (BZ#1668762)
• Previously, new installations using Heketi had tuned the self-provisioned heketidbstorage gluster volumes for db workloads but did nothing to volumes created by older installs. Now, when a container hosting Heketi is first started and a built-in tool attempts to automatically tune the settings of an existing heketidbstorage volume. (BZ#1756861)
• With this update, a new environment variable HEKETI_LVM_WRAPPER has been added. This environment variable enables Heketi to call a wrapper script inside the RHGS-Server containers to execute LVM commands. This improves stability of LVM initialization and handling of any LVM events during runtime. Race conditions between the LVM stack on the host and inside the container do not occur. Now, the LVM commands are executed through a wrapper script that runs LVM commands on the host instead of inside the container. (BZ#1772612)
• Previously, a condition existed within Heketi where an error running a command on a node failed and was being cleaned up for a volume expand operation that would lead to incorrect used/free space on the devices related to that volume. With this release, this issue has been corrected and Heketi now correctly updates the free space even when the `expand` operation fails. (BZ#1653567)
• Previously, Heketi was not consistent about how the status of devices was reported through the Heketi-cli tool. Now, Heketi always reports `removed` rather than `failed` when reporting that state. (BZ#1710685)
• Previously, Heketi could crash if an attempt was made to delete a brick from a node that was not running and the option to `debug unmount` failures was enabled. Heketi no longer crashes when this condition occurs. (BZ#1796059)

Users of OpenShift Container Storage 3.11 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 Gluster Storage Server for On-premise 3 for RHEL 7 x86_64

Fixes

• BZ - 1653567 - Incorrect used and free space shown in heketi topology info
• BZ - 1710685 - After a device is removed from heketi, status of the device is different in device info and node info
• BZ - 1718982 - Heketi spec contains old reference to image tag
• BZ - 1756861 - Automatically tune heketidbstorage for safer workloads (on container start)
• BZ - 1772612 - heketi should call the wrapper LVM scripts instead of the lvm binaries in the gluster container
• BZ - 1789029 - Gluster pod going into "0/1" state after some tests execution
• BZ - 1790560 - [OCS-3.11.z][TRACKING TICKET] gluster-block-target.service not starting after a reboot with a genconfig failure
• BZ - 1796059 - Heketi crashes if the pod is not running on the node during brick remove

CVEs

(none)

References

(none)