Reinstall from scratch, old disk was not cleaned up: cockroachdb ends up in weird state


I want to only mention following.

We had cockroachDB in use.
We removed it and reinstalled it.

We did not clean up / remove the disk between those steps, before installing again. So, there was “garbage” files on disk from the previous installation.
It would be possible configure removal of the disk as automatic, when helm delete is executed. Now, the old disk was reused when CRDB was installed again.

Inserts worked.
Following operations failed: truncate (table), create table.

Failure message was: " pq: split failed while applying backpressure: could not find valid split key"

I do not know should CockroachDB be able to survive / recognize such situation somehow. I do not request such a feature. Again, I am only mentioning such situation.

We need to clean up / remove the old DB disks in k8s, after removal of CRDB (helm delete) and before CRDB would be installed again (helm install).
The situation is not available anymore in our system.

Hey @roachman,

Thanks for letting us know.

What version of CRDB are you running?