-
Notifications
You must be signed in to change notification settings - Fork 42
Description
Hi Quickwit team,
I'm running Quickwit on a Kubernetes environment with indexes stored in an S3 bucket. During the initial deployment, the index creation works as expected.
However, when I upgrade my environment (via Helm or otherwise), the Quickwit pod fails with an error indicating that the index already exists.
Here’s the command I’m using inside the pod:
`bash
Copy
Edit
/bin/bash -c "quickwit source describe --index {{ .index }} --source {{ .source.source_id }} --endpoint ${QW_CLUSTER_ENDPOINT} || quickwit source create --index {{ .index }} --source-config {{ .source.source_id }}.yaml --endpoint ${QW_CLUSTER_ENDPOINT}"
This logic assumes that if the source doesn’t exist, it should create it—but in case the index already exists, Quickwit still fails. It seems like the source create or something internally is checking for the index even before the source creation.
`
Expected behavior:
When upgrading the environment, if the index already exists, Quickwit should not fail but continue gracefully.
Alternatively, the CLI should provide a way to check and skip creation if the index already exists.
Suggestion:
Add idempotency or a built-in flag like --skip-if-exists for quickwit index create or quickwit source create.
Or provide a command to safely check and conditionally create both the index and source.
Let me know if I can provide any logs or more details to help reproduce the issue.
Thanks for the great work on Quickwit!