DRBD verify-alg mismatch after LINSTOR/Piraeus upgrade due to runtime/config inconsistency

[saarthak18]

Description

We observed DRBD connection issues after upgrading Piraeus/LINSTOR container images, without changing the Helm chart version.

Upgrade Details

We did not upgrade the Piraeus Operator Helm chart.
The chart version remained the same.

However, we upgraded the following images:

piraeusdatastore/piraeus-server → v1.33.1 (from older version)

piraeusdatastore/piraeus-csi → v1.10.5 (from older version)

piraeusdatastore/piraeus-ha-controller → v1.3.2 (from older version)

Observed Issue

After the image upgrade, some DRBD resources went into a disconnected state with the following error:

Different verify-alg settings. me="sha256" peer="crct10dif"

Resulting behavior:

DRBD connections transition to StandAlone

Replication breaks

PVC becomes degraded

Investigation

We observed:

LINSTOR configuration:
DrbdOptions/auto-verify-alg = sha256
DRBD runtime (drbdsetup show):

Some nodes → verify-alg = sha256

Some nodes → verify-alg = crct10dif

This indicates a mismatch between desired config and runtime state.

Questions

Is this behavior expected when only upgrading container images (without changing Helm chart)?

Should existing DRBD resources be automatically reconciled to avoid runtime mismatch?

Could this be caused by a race condition during resource reconciliation?

Is there a recommended upgrade procedure to avoid this issue?

[saarthak18]

@rck can you comment on this