PhanLe1010

Lot of good statistics at https://github.com/longhorn/longhorn/issues/8803#issuecomment-2197297624 @ejweber 🚀 :D > In total, that's a 55.5% chance of AT LEAST a 10 second delay (and a 55.5% * 44.4% = 24.6%...

> I was thinking of another idea @james-munson and @PhanLe1010. I can't remember if I said it during the workshop or not. > > In the fast failover case, we...

🚀 That is a great statistic calculation! Many thanks!

As discussed with @james-munson, currently Kubernetes has 4 access modes in total ([link](https://kubernetes.io/docs/concepts/storage/persistent-volumes/#:~:text=In%20the%20CLI,RWOP%20%2D%20ReadWriteOncePod)): 1. RWO - ReadWriteOnce 1. ROX - ReadOnlyMany 1. RWX - ReadWriteMany 1. RWOP - ReadWriteOncePod We...

> simultaneously. For example: > > ReadOnlyMany, and > ReadWriteMany. > Then, we may be asked to publish the volume with either of these access modes. > > What should...

> Instead of "only" doing this on the mount side though I believe we'll want to change the actual export to be read-only (or rather, not rw), so that the...

I think a backoff mechanism makes sense when there is a non-recoverable error. For the type of backoff. Even though I agree that a fixed backoff period is straightforward, it...

I personally think that we should wait for the complete solution instead of merging temporary solution because it would require document change, user experience change later on. WDYT @jangseon-ryu ?...

Hi @jangseon-ryu If you have a patch commit, I can help to review it and you can use it for your cluster. However, I think it is not suitable to...

> Dec 19 13:19:36 k3s-node-2 iscsid[3160778]: connect to 10.42.3.235:3260 failed (No route to host) > There is no such ip in k3s Looks like the orphan tgt target leftover after...