Hello, I would like to understand something that happened to me recently.

I have cinder configured with glusterfs and I had a volume that I couldn't attach, after watching logs and the like I noticed that it was a split brain situation but anyway, this is a replicated distributed volume and I assume that in the case one brick failed cinder would be able to work out using brick's replica.

Doesn't this work with split brain conditions? I mean, if it wasn't for the split brain issue, cinder would've been able to use the replica when the other brick failed?

Thanks in advance!