I am running OCFSv2 in production in 2 month, and in staging for approx
6
month. While it is stable in _normal_ conditions, failure tests shows,
that
it don't survive many failure scenarios (causing all servers to reboot,
or
switching to readonly mode). So, it is not a simple choice - use it or
not.

...Alexei is right. When OCFS is not challenged with abnormal conditions
to respond to it is largely just fine. But then cluster software is
supposed to be able to handle edge cases--that's why we cluster. In our
closest estimate here at PolyServe, over 50% of all of our code only
executes in cases where servers are not behaving corectly for any of
such reasons as interconnect/SAN/host responsiveness issues followed by
the transition/recovery and fencing that must occur. Those are the hard
parts to get right in clusterware...unfortunately you can't "benchmark"
such "features"

...A parallel to this is shared-nothing database architecture. If you
code apps correctly and tune it right and don't have any stability
issues in production, such applications can "work just fine"...the wheat
seperated from the chaff as soon as nodes start to flake out

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