We recently installed Red Hat Storage (RHS) on a cluster that had previously been running GlusterFS 3.3.1. A lot of paths are now inaccessible via NFS at the client, mainly due to metadata split-brain errors which are reported in the logs. In some cases this prevents access to the whole volume or large parts of a volume.
The GlusterFS client mount points are now prone to hanging, and the only solution in some cases is to reboot the client machine. I suspect this is related to the split-brain problems that is causing RHS to block access to affected paths.
To solve both these problems I need a way to fix all the metadata split-brain errors automatically. I know how to do it manually, using setfattr, but it would be completely impractical to do that for all affected paths.
Red Hat Storage Server - 2.0
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase of over 48,000 articles and solutions.