[Question] weaviate pod restart unavailable - no locks available

Support
1

the weaviate pod does not go up again, so when I look at the log,
I see the error message “error”: "init bolt_db: open "/“var/lib/weaviate/schema.db": no locks available”.

This occurs when the node with the pod up is forcibly stopped and displayed for availability test, and even if all the weather pods are deleted or Helm uninstalled with the existing values.yaml setting and then installed again, the same error occurs and the pod does not run normally. (status: CrashLoopBackOff)

How to solve it?

2

the weaviate was running in k8s cluster with two nodes. and the schema.db file was in NFS(BlockStorage).
And I can still access the path of schema.db through node server and the schema.db, tx.db file and other files and collections directory.
But it still shows ‘no locks available’ error.

And I tried to make new weaviate instance, but it also shows no locks available error and there’s only 0 size schema.db file.

3

Hi!

I believe this might be something with how you are mounting the volume to Weaviate.

If you use regular volumes, not NFS, do you still see this kind of issue?

4

We’ve solved it by changing NFS mount options from ver3 to 4.
ver3 is using lockd, statd to lock files but I think it’s not that stable.

1 Like