Shuo Wu

This feature is related to the manager repo only: https://github.com/longhorn/longhorn-manager There are 2 parts you can refer to before starting: 1. Pick up a random setting and see how other...

Make sense! 1. We can verify if the system snapshot mismatching would lead to data inconsistency: After the rebuilding & snapshot prune, recreate one more snapshot then retry purging the...

I just found one more solution that can avoid the inconsistency without using a lock: 1. Read the volume head fiemap and get the data extent info from one random...

1. It's about how the rebuilding can be improved. Currently, Longhorn will try to reuse the failed replica if possible. But it will read and verify all existing data for...

Can you provide more details about the reproducing step...?

> I tried to manually perform the backup by clicking on "run now" in the options of the cronjob: a pod is created to handle the job and I checked...

I saw tons of the logs like the followings: ``` 2022-08-07T15:44:51.547638859+02:00 time="2022-08-07T13:44:51Z" level=warning msg="Cannot take snapshot backup" backup=backup-989d2d761c4844c3 controller=longhorn-backup error="could not find snapshot 'sched-ba-353e2804-4e70-48b3-8f39-a91d375c366e' to backup, engine 'unifi-controller-vol-e-36c0bc0d'" node=nem-03 2022-08-07T15:44:51.547722181+02:00...

Can you try to remove all stuck backups then see if the backup creation works?

You can check the backup state in GUI or `kubectl`. If a backup is an old one and the state is not `Completed`, you can consider it as a stuck...

EXT4 is more stable, at least for Longhorn volumes. It won't be deprecated. But YES we can update the doc. cc @innobead