I made an experiment , using OCFSv2 for document storage in development
(document storage is for the customer's documents, so it keeps thousands of
files).

Results:
- 2 full cluster failures in 2 weeks;
- 1 failure - local file system was damaged (I guess, OCFS wrote wrong
buffer in the memory) so I need to reinstall 1 node;
- second failure - 1 server freeze and another could npot get OCFSv2
resource.
- post-morten analyze showed 1 slightly damaged dierctory (# of counters was
wrong).
- no data losses.
- performance is average - sometimes o2cb spent a lot of CPU.
- works faster vs NFS (NFS on NetApp, OCFSv2 on NetAPp / iSCSI).

kernel - 257 (SLES9 SP3 + update, i386).

Most annoying problems:
- overall stability - I saw many times (including this experiment) how
system self-panicked because of _fencing_.
Notice that SLES9 have 'no reboot on panic_ default (btw, why SERVER have
not REBOOT ON PANIC by default). I'd like _in many cass_ specify _in case of
OCFSv2 problems - just unmount FS but dont reboot_ - no way.

- do not support multiple heartbeats. I use 3 heartbits (2 ethernets + 1
serial in heartbeat), and I cannot use second interface with OCFSv2 -
noncence!

- System reboot require special protection which _GARANTEE_ that system
reboots even f some process can not be stopped.

- if symlink is not resolved, it is reported to syslog - no need, just a
noise.

Conclusion - negative for document storage (for now).

Does anyone have opposite results (not with oracle DB or backups - they do
work in my lab without any problems)?




--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting