BsoBird commented on PR #9546:
URL: https://github.com/apache/iceberg/pull/9546#issuecomment-1926106692
If the user's fs.rename call is taking too long (slow file system response)
and the user has set the metadata-ttl to be too short, then it is possible that
the user will commit an older version of the metadata successfully.
Suppose client A calls fs.rename, and the current client is ready to commit
v2.
However, because of the slow response of the file system, it takes 30
seconds for the operation to complete.
At this point, there is another client B that is committing every second.
And the TTL for this table is only 2 versions.
When client A calls successfully, assuming that the maximum version has been
rolled over to V60, and there are only two versions of metadata, v60 and v59,
then fs.rename(v2) will succeed because v2 no longer exists and there is no
conflict.
Therefore, the final result is: the table retains all three versions of
metadata v60, v59, v2.
When this happens, is it acceptable?
@RussellSpitzer
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org
For additional commands, e-mail: issues-h...@iceberg.apache.org
