Closed tehsphinx closed 3 days ago
In case we should just update nats-server to fix the issue, kindly let us know (but I'll try that next).
Since the 2.6.6 release is nearly three years old at this point, we would indeed normally ask if it's still reproducible on the latest version.
I'll look into that and report back. Thank you!
Updated to v2.10.22-alpine3.20
.
Unfortunately that development environment still had a version where the JetStream data folder was not mounted. Updating the docker image wiped all data.
We'll have to get back to you on this once the disk runs full again. Until then I'll close the issue.
Observed behavior
What we observe when the stream / KV bucket is broken:
Same for
del
orpurge
operation (and I assume any write operation):Expected behavior
Optimally a restart of the NATS server would restore the KV bucket.
Alternatively we'd need a way to restore the stream manually.
We already tried to backup -> delete -> restore the underlying KV_bucket stream, but that doesn't fix the issue.
Server and client version
Client: nats-cli (devel based off commit 4da881fdf5c87d6b064d6f8d2eb936f3802f4f73 from 8th Oct 2024) Server: nats-server: v2.6.6
Host environment
No response
Steps to reproduce
After the disk ran full (a few times) on our development environment we observed that some stream are broken. In this ticket the stream is the one powering a KV bucket. (The full disk is due to another application, not NATS JetStream).
Reproduction steps: