[DRBD-user] Zombie, Zombie?

Andreas Semt as at computer-leipzig.de
Tue Apr 27 12:41:22 CEST 2004

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


Andreas Huck wrote:
> Hi,
> 
> On Monday 26 April 2004 22:10, Andreas Semt wrote:
> [...]
> 
>>Right now i had the same phenomenon: A load average of 16 (!) and cpu
>>almost idle. No (read/write) access to the partition on the drbd device
>>possible!
>>Can i see what causes that "hang" i.e. the processes who read/write do
>>that partition/drbd device?
>>Can i enable a debug mode of drbd to see if drbd has something to do
>>with that?
>>Can I see some kernel messages regarding disc access?
>>
>>THANKS for help! That problem is REALLY annoying!
> 
> 
> have you monitored the disk IO load (e.g. vmstat 1)? If it stays high
> than maybe your RAID is trying to catch up - and your system waits 
> for it patiently. 
> Does disabling the hardware RAID cache change the behavior?
> 

No, at this moment I did not monitored with vmstat. The problem is, that 
the "hang" occurs at random, so i have to wait or simulate it to 
monitor. Also I don't believe it's the RAID write cache, because the 
system and the drbd devices are located on the same RAID 5 array and i 
had no problems to access the system partition during the "hang".

Are there any tools to generate disk load on a particular block device 
and to monitor read/write precesses on that device (like an "i/o test 
suite")?

Other hints (perhaps how to make drbd more communicative)?

-- 
Best regards,
Andreas Semt



More information about the drbd-user mailing list