Open seatunneltrailuser opened 8 months ago
error while getting created time for file "919-sgnl1jjo.parquet":creation time is not available for the filesystem
@seatunneltrailuser What are you using as the filesystem? Could you please describe your setup?
I have seen the same error on my tests. My host file system is xfs.
what is the status of this? Will this be fixed by Cube? Is there a List of supported Filesystems?
I've consulted with you-know-who and it looks like the root cause is the lack of the "creation time" support in XFS. I guess the workaround will be either to try -m crc=1
or switch to another filesystem:
@thePermission I guess any file system the stores a file "creation time" (e.g., ext4
) should be OK. I'm going to add a note in the docs but it's outside of a roadmap to get this fixed. Contributions are always welcome, though!
I would also add that using CUBESTORE_REMOTE_DIR
is not recommended in production, so this whole issue should not be a huge blocker for production deployments:
Describe the bug While the Refresh Worker is performing pre-aggregations, we get multiple of this error
ERROR [cubestore::remotefs::cleanup] error while getting created time for file "919-sgnl1jjo.parquet":creation time is not available for the filesystem
For every intermediary file created by the pre-aggregation process, CUBESTORE is also performing cleanup of these files. It is failing to purge these files due to this issue - Creation time is not available for the filesystem.
To Reproduce Steps to reproduce the behavior:
Expected behavior These files are successfully cleaned up by the [cubestore::remotefs::cleanup] process.
Version: Docker image cube/cubestore:latest
Additional context Is it safe to clear these files manually ? Also, Are we missing any additional parameters to be set while starting up the RefreshWorker the job ?