[CTDB and GFS] test: scenario and some results

[tridge at samba.org]

Ignacio,

 > 1. The LB was a bit a mess, basically because the ServerIron I used
 > can't test the samba at L7 level (not in the ServerIron's well know
 > ports). So, if the eth0 interface is up, the traffic is balanced also to
 > the stopped samba, and the things goes very badly.
 > 
 > However, if I perform an "ifdown eth0" in one of the samba nodes, the LB
 > goes very well.

You can bring the public interface up/down using ctdb event
scripts. The default event scripts just add/remove IPs, but leave the
interface up. You could either modify /etc/ctdb/events.d/10.interface
or create a new event script which brings the interface down
completely if it is no longer serving any IPs. 

If you have any problems with the scripting to do this let me know and
I will give you an example (basically count the number of IPs on the
interface using /sbin/ip and wc, then bring it down if its zero)

 > EXT3 samba-3.24
 > ---------------
 > 1-node ops average = 8619 ops
 > 
 > GFS
 > --------------------------------
 > 1-node ops average = 7401 ops
 > 2-node ops average =  204 ops
 > 3-node ops average =  469 ops

I should note that with "posix locking = no", the RAW-BENCH-LOCK
script is testing CTDB and your network, not GFS. You should see close
to zero GFS filesystem operations with that test.

Basically the RAW-BENCH-LOCK test is rather like the ping_pong test,
except over the SMB/CIFS protocol rather than to the filesystem.

Cheers, Tridge