2008/7/21 Etienne Goyer <[EMAIL PROTECTED]>: > That is interesting. Can you expand a bit on the stale file handle > problem you had with NFS? Was a public bug ever reported on the > subject? Is it something that is easy to reproduce? If not, can you > give a hint about the conditions that would trigger this problem?
It's not really a bug. It's a fundamental design problem with NFS and clustering. NFS creates its file handles from the inode number of the file on the server. If you have clustered servers then you must make sure that the shared filesystem is *identical* at the inode level - and that generally means that any file mirroring you do must be at filesystem level, not file level. (DRBD generally or a posh SAN). If you don't then as soon as the server changes identity, client NFS will issue the dreaded 'stale file handle' error message and you're out of action until you remount the share. NFS doesn't appear to have an effective client side recovery on a cluster in Ubuntu. Or if it does it isn't on by default. NeilW -- Neil Wilson -- ubuntu-server mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-server More info: https://wiki.ubuntu.com/ServerTeam
