On 2014-08-14T18:02:05, steve
1. Where our single ext4 file server is predictable under load (it just gets slower), with both nodes up, why does the cluster fail so badly under ext4 but absolutely screams under ocfs2? The strange thing is that the clustered ext4 actually performs better when only one node is up. ext4 cannot be concurrently mounted on multiple nodes. Hi. Other setups use it: http://sigterm.sh/2014/02/highly-available-nfs-cluster-on-debian-wheezy/
But that's not what they are doing. They are only mounting it once; with drbd (in normal configuration) it's only possible to write to the device once anyway, so no-concurrent mount would be possible anyway. (They are also using a very old heartbeat-v1-style fail-over, not a more modern pacemaker setup.)
We proved it doesn't work well and it usually freezes hard needing to hold down the power off button.
This would be a bug. This shouldn't happen.
The file system just plain-out does not support that; if that is required in your use case, you have no choice but to use OCFS2 (or try GFS2). That's fine. We just wondered if there was a simple one liner which explained this. That it just doesn't work is OK. But some people need convincing.
The simple one liner is that ext4 is a local file system (and thus suitable only for local or fail-over style use in a cluster), and OCFS2 is a concurrent cluster file system. Regards, Lars -- Architect Storage/HA SUSE LINUX Products GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) "Experience is the name everyone gives to their mistakes." -- Oscar Wilde -- To unsubscribe, e-mail: opensuse-ha+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-ha+owner@opensuse.org