Weekly fstrim and mkfs discard operation hangs on DRBD SSD device

This document (000019631) is provided subject to the disclaimer at the end of this document.

Environment

SUSE Linux Enterprise High Availability Extension 12 SP3

Situation

When running a command that triggers a BLKDISCARD on an SSD device used for DRBD backing storage, such as the weekly fstrim.timer or a mkfs command, the operation hangs. Write IO to the device hits 100% utilization for its duration to write over the device. This can result in a monitor failover of a DRBD-backed Filesystem resource or failures of remote NFS client mounts that are dependent on data stored on the DRBD device.

Resolution

There are two options to work around the issue:

Upgrade to drbd-9.0.14+git.62f906cf-3.20.1.x86_64.rpm.
Perform a Service Pack migration to SLES12 SP4 or newer.

Cause

Previous DRBD versions in SLES12 SP3 used WRITE_ZEROS for DRBD trim. The bug fix uses WRITE_ZEROS when the lower IO layer cannot handle the trim, otherwise, a standard TRIM is run against the block device.

Disclaimer

This Support Knowledgebase provides a valuable tool for SUSE customers and parties interested in our products and solutions to acquire information, ideas and learn from one another. Materials are provided for informational, personal or non-commercial use within your organization and are presented "AS IS" WITHOUT WARRANTY OF ANY KIND.