You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At least, the behaviour should be consistent. I presume minSuccess should never be negative but haven't researched what it should be.
At a higher level, I don't think an absence of ingesters should cause queries to fail. We can tell, without asking, that the empty set of ingesters holds an empty set of chunks, so the query should return whatever is in the persistent store.
The text was updated successfully, but these errors were encountered:
I hit this recently testing the Cassandra branch. Have opened #681 to resolve it IMO.
At a higher level, I don't think an absence of ingesters should cause queries to fail. We can tell, without asking, that the empty set of ingesters holds an empty set of chunks, so the query should return whatever is in the persistent store.
That sounds useful for other things, but this issue mainly arises from misconfiguration, right? In this case I think automatically only returning data from the chunk store might be a bit confusing. In normal operation, the ring should never be empty. Having a Cortex mode that only ran from chunk store (read-only) does sound cool though.
This issue has been automatically marked as stale because it has not had any activity in the past 30 days. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.
While testing, I had zero ingesters owning tokens. In this situation, Distributor.Query will:
queryIngesters()
which will compute a negative number as the minimum required for success, hence return success with no data.At least, the behaviour should be consistent. I presume
minSuccess
should never be negative but haven't researched what it should be.At a higher level, I don't think an absence of ingesters should cause queries to fail. We can tell, without asking, that the empty set of ingesters holds an empty set of chunks, so the query should return whatever is in the persistent store.
The text was updated successfully, but these errors were encountered: