Shuo Wu
Shuo Wu
Is this reproducible in v1.1.2? IMO, it's unlikely a regression... If it's not a regression/it's reproducible in v1.1.2, the PR should be backported to v1.1.3.
One alternative in my mind is that we can launch a new CRD similar to VolumeAttachment. Then all attachment/detachment-related info(e.g., if the volume is attached by the system) would be...
Some suggestions for further testing and investigation: 1. Use version v1.2.4 to avoid the potential impact of feature snapshot pruning (which is available since v1.2.5). 2. Does Longhorn need to...
When all replicas are crashed, the volume will be detached & faulted first. Sending salvage requests during the detaching leads to this failure. I will check https://github.com/longhorn/longhorn/issues/4383 later.
cc @joshimoo
Probably it can be updated and stored in `snapshot.Status` but with an async way.
@iosifnicolae2 Volume `pvc-61defda0-969a-4404-a076-1ce0de63b791` starts the snapshot surge(actual cleanup) but there are replica eviction and volume detachment happening simultaneously around `2022-01-24T20:40:21Z` Detachment log: ``` 2022-01-24T21:40:25.164278827+01:00 time="2022-01-24T20:40:25Z" level=debug msg="Volume pvc-61defda0-969a-4404-a076-1ce0de63b791 unpublished from...
@bagdus No. Longhorn is a block-level storage system hence it does not understand the filesystem. This is the key point for your question. In your/filesystem perspective, the space will be...
@iosifnicolae2 Deleting any snapshots will trigger system snapshot cleanup.
> @shuo-wu I've deleted the latest snapshot created manually, but I still have a system snapshot on my disk even though on Longhorn UI it says that it was deleted.....