[DRBD-user] XFS issues

sebastian schmitzdorff sebastian.schmitzdorff at ina-germany.de
Mon Apr 11 08:49:57 CEST 2005

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


> Last night I just completed a two-node cluster running Debian GNU/Linux
> Woody with some backports from Backports.org to allow it to run Linux
> kernel 2.6.11.6 and DRBD 0.7.10. The machines are IBM xSeries 345, with
> 4 x 2.8GHz logical Xeon processors, 2GB RAM, and 4 x 36.4GB 10,000RPM
> SCSI hard drives in RAID 5EE. They are connected through their e1000
> network interfaces by a crossover cable, with MTU for both interfaces
> set to 9018.
> 
> The system seems to be fine. Initial synchronization went well, failover
> works fine, and so far the filesystem has not yet experienced any
> corruption. Last night I ran bonnie++, and will find out later in the
> day how the system fares. I'm running bonnie++ not as much to test
> performance but to test how XFS + DRBD will respond under the load it
> puts.
> 
> Can those who have tested the XFS + DRBD combination and gotten into
> problems share what they did that led to these problems? I'd like to try
> to reproduce all these and see how the cluster fares. I will let the
> list know, of course.
> 
> Thanks a lot in advance.
> 
>  --> Jijo
> 

Hello Frederico,

to crash my XFS + DRBD system it only needs some time and a moderate
load, there are approximatley 500MB of log files written every day. It
takes roundabout 1-2 day's here until it will crash during
sychronization.

I will seriously consider moving from XFS to ext3 the follwoing days,
since I have no indication that this issue will be addressed soon
(unless I buy a support contract for drbd, which i cannot do).
I'm quite interested how your testing goes.

greets

Sebastian




More information about the drbd-user mailing list