[Linuxha-users] Re: LinuxHA failover problems

[Simon Edwards]

On Mon, 2005-01-24 at 21:42 -0400, James MacLean wrote:
> Looks like it's completely resyncing. On the plus side, no errors in 
> clstat that I could see ;) :

Ok ... after more testing - below is the resync of a failed node joining
the cluster ;

[root at rserver2 ~]# head /proc/drbd
version: 0.7.6 (api:77/proto:74)
SVN Revision: 1663 build by root at rserver2, 2005-01-22 22:50:56
 0: cs:SyncSource st:Primary/Secondary ld:Consistent
    ns:13420 nr:116 dw:11824 dr:14580 al:12 bm:3 lo:83 pe:74 ua:106 ap:0
        [========>...........] sync'ed: 44.5% (20508/33636)K
        finish: 0:00:01 speed: 13,128 (13,128) K/sec
 1: cs:Unconfigured
 2: cs:Unconfigured
 3: cs:Unconfigured
 4: cs:Unconfigured

Notice that is is only having to sync 33636Kb - the file system size is
2Gb - so it is performing a partial resync.

[root at rserver2 ~]# df /apache
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/drbd0             2088692     12812   2075880   1% /apache

Now - for testing if you wish to simulate a node failure then doing a
"reboot" or similar, or even pushing the off button if configured for a
"soft reset" will cause a complete resync to take place...

This is because unless you drop the power to the node it will still
attempt to close files etc - which will update drbd on that node - now
both nodes will have different generation counters ... and hence a full
sync must be performed for the sake of sanity.

Hence if you've not done so please re-try the tests - this time pulling
the power cable or switching of the PSU at the rear of the machine ...
you get the idea! Let me know if this works for you.

Regards,
Simon.