0.4.2
VolumeCommitError
Package: flyteplugins.union.errors
A keep-alive checkpoint could not be made durable: the writeback staging queue did not finish uploading to object storage (it timed out, or its progress couldn’t be observed because the mount was wedged/gone), so publishing the snapshot would seal a metadata index that references chunks not yet in object storage. A later hard-kill would then lose those chunks, leaving dangling references that read as EIO. The live mount is intact and its data is not lost — retry the checkpoint once uploads catch up.
Parameters
class VolumeCommitError(
message: str,
)| Parameter | Type | Description |
|---|---|---|
message |
str |