Segmentation Fault about longhorn HOT 4 CLOSED

Can you list the reproduce steps?

from longhorn.

On digging further, found following :

• If I deploy manager in namespace1 (in which one volume was previously attached), it runs perfectly with following logs :

INFO[0000] Detected IP is 10.32.0.12
INFO[0000] Kubernetes orchestrator is ready
INFO[0002] Listening on 10.32.0.12:9500
DEBU[0002] Created volume by spec my-volume-1
DEBU[0003] Start managing volume my-volume-1
DEBU[0628] volume my-volume-1 desire state is detached
DEBU[0628] volume my-volume-1 state is detached
DEBU[0632] volume my-volume-1 state is detached
DEBU[0632] volume my-volume-1 desire state is detached
DEBU[0632] volume my-volume-1 state is detached
DEBU[0633] volume my-volume-1 state is detached
DEBU[0633] volume my-volume-1 desire state is detached
DEBU[0633] volume my-volume-1 state is detached

The logs are continuous.

• If I deploy it in any new namespace, the segmentation fault occurs.
• No pods are currently requesting PVC / flexvolume named "my-volume-1". All pods are deleted. Still the manager is having logs related to "my-volume-1" which was previously used.
• Is it possible that manager is saving these volume names somewhere ? I checked all hostPath directories but have found nothing yet.

from longhorn.

Sounds like CRD wasn't cleaned up.

I will include that in the README.md as well.

For now, can you use kubectl -n longhorn-system delete crd -l longhorn-manager to clean up the CRD stored in the Kubernetes and try again?

from longhorn.

Wonderful.
Confirmed, working now in separate namespace.
Closing. Many thanks for the support.

from longhorn.