Shuo Wu

> Thank you for the reply @shuo-wu I thought i understood this, but i apparently do not properly understand it. I have a snapshot with a very large indicated size,...

> How do you think about this. We have a backoff limit so we don't mark the rebuilding replica as failed too fast though. We already introduced the backoff interval...

This seems to be a Rancher issue rather than a Longhorn issue: https://github.com/rancher/rancher/issues/34688

The Longhorn version is v1.2.4. The direct reason of the volume getting stuck and the only available replica (the gray one in the screenshot) cannot become mode `RW` is, the...

OK. My last comment & analysis is for the original issue. I will check the comment you mentioned later.

Hey team! Please [add your planning poker estimate](https://app.zenhub.com/workspaces/Longhorn-Sprint-59d8105b2fbdf4155c4589ef/issues/longhorn/longhorn/4366?planning-poker) with Zenhub @derekbit @ejweber @PhanLe1010

Some problems in this PR: 1. The seq-related bench may be meaningless. If multiple threads are enabled during a seq bench, `seq` means each thread is responsible for operating its...

The unit test involves 1~10G data IO hence it would be time consuming in relatively low spec env. The modification you mentioned looks good.

Probably NO. Should we mention unexpected Longhorn bugs?

Actually, in case of unclean unmount triggered by (IM pod crash or OS reboot/...), I am thinking if mount operation will replay the journal for the filesystem automatically. If NO,...