[DRBD-user] DRBD serious locking due to TOE - UPDATE

Carson Gaspar carson at taltos.org
Sat Jan 5 07:23:56 CET 2008

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

Ben Clewett wrote:
> For those of you having locking and following this thread, I have found 
> another problem related to our HP Proliant servers.
> These optionally come with a battery backed up write-ahead-cache on the 
> cciss controller.  I have 256M of write ahead cache active.
> I found this thread from Lars suggesting these items may have a problem 
> where they miss-report storage of data:
> http://lists.linbit.com/pipermail/drbd-user/2007-June/007033.html

There is no problem, as long as your write cache is battery backed. If 
it isn't, you're risking your data. But that isn't your performance 
problem (it makes things faster, not slower...). It may, however, make 
your performance uneven. Increasing your buffers allows drbd to queue 
more data, and smooths out the cciss driver's response time impact.


More information about the drbd-user mailing list