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
The issue is that when we request the earliest offset from the broker (ListOffsets request type / beginning_offsets in Python)), it returns the earliest available locally, i.e. not counting remote.
The text was updated successfully, but these errors were encountered:
@mdedetrich did the patch work for you? It doesn't for me, the kcat test.
At least log.logStartOffset as a parameter to findOffsetByTimestamp will not produce anything, because log.logStartOffset is always (currently) higher than anything in the remote storage.
Also, based on how oldStyleOffsets is used later, it should always be 0 or 1 element, so I guess ++ won't work.
Indeed this is correct, I was trying to write a test to validate this and I noticed the same. I was attempting to upstream this first and the core implementation of ListOffsets has changed compared to whats in 3.3.
Assuming the beginning of the log is on remote
fetches only the locally available records, but
fetches records from the remote storage.
This Python
fetches from the remote.
The issue is that when we request the earliest offset from the broker (
ListOffsets
request type /beginning_offsets
in Python)), it returns the earliest available locally, i.e. not counting remote.The text was updated successfully, but these errors were encountered: