[DRBD-user] DRBD deadly slow

Florian Hackenberger f.hackenberger at chello.at
Thu Nov 13 17:55:35 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.

On Thursday 13 November 2008, Felix Zachlod wrote:
> Trying to write to
> the drbd1 device gives a performance of ~ 3MByte/s if both nodes are
> online and ~8Mbyte if writing to the primary node where the secondary
> is offline. I watched the webinar about performance tuning for drbd
> and tried several options such as max-buffers, al-extents and
> no-disk-flushes which all in all did not help at all. Performance
> increases with this tuning parameters are not more than 10-15 percent
> whereby this might be measuring inaccuracy too.

Resending to get the thread id correct, my previous mail was sent 
without hitting reply because I could not find the mail I replied to.

Dear Felix,

Please make absolutely sure that you are running a version of DRBD which 
has the options no-disk-flushes and no-md-flushes actually built in. 
Without these options being active, DRBD waits until the secondary node 
has flushed the write request to DISK before finishing the write 
request on the primary node and returning to the application which made 
the request. Please note that setting the Protocol to A or B did not 
have any effect on performance on my setup (8.0.11 on Ubuntu Hardy).

See http://archives.free.net.ph/message/20081023.153142.9b517971.en.html


DI Florian Hackenberger
florian at hackenberger.at

More information about the drbd-user mailing list