All,
I am trying to optimize GFS lock performance and using the CTDB
tool "ping_pong" to test the number of locks per second on my GFS
filesystem. I am consistently seeing between 95-98 locks per second.
According to CTDB, a good clustered filesystem should support 20k+ on a
single node.
I found that there is a default limit of 100/sec defined by the
gfs_controld daemon. According to the man page for gfs_controld I should
be able to add an entry to /etc/cluster/cluster.conf to adjust this
limit:
<gfs_controld plock_rate_limit="100"/>
I have tried may variations of adding this entry to my cluster.conf, but
in all cases system-config-cluster barfs on syntax. In addition, the
gfs_controld can have this option set at boot time via the "-l<N>"
command line option, however, there is no facility for changing this vi
the init script.
Is anyone else adjusting this parameter? If so, how are you doing it?
Thanks,
Kevin
_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list
