[DRBD-user] Slow syncall after kernel+drbd upgrade

Francis SOUYRI francis.souyri at apec.fr
Thu Jan 22 08:45:51 CET 2004

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


Hello Christian,

Christian Hammers wrote:

>Hello
>
>On Wed, Jan 21, 2004 at 01:34:31PM +0100, Lars Ellenberg wrote:
>  
>
>>>Hm, as I have to reboot the machines anyway due to a problem 
>>>with a NIC (which is not involved in DRBD), I will probably rather
>>>downgrade to 0.6.4 and see what happens there.
>>>      
>>>
>>yes, you are right, this would be nice to know. though I doubt
>>that it makes a difference.
>>anyways, "never change more than one thing at a time" :)
>>    
>>
>
>The third night spending my time with this !"§% system... 
>
>First, I switched back to 0.6.4 but the syncall speed stayed slow
>so I disconnected the server and rebootet 0.6.10 again.
>
>Then I commented out the ts-size=5000 parameter to reset it to the 
>default of 256. Didn't gained me much, the speed even dropped to
>about 1100 KB/s :-(
>
>So the only things that could have an influence and were not yet tried
>are:
> - Kernel change from 2.4.23 to 2.4.24 due to the security issue
>  
>
Vanilla kernel ? Distrib kernel ?

> - New CPU+Mainboards (Intel P4 & AMD-Athlon, both about 2,6GHz)
> - New internal NICs, both Intel Gigabit with e1000.o driver
>  
>
Could you create a ram filesystems on each node and start a rsync.
Do you have the possibility to test other NICs ?

>As somebody asked me for it I have the iostat and vmstat lines of both
>systems here:
>
>[the primary]
>
>Device:            tps   Blk_read/s   Blk_wrtn/s   Blk_read   Blk_wrtn
>dev8-0           58.00      6840.00         8.00       6840          8
>
>   procs                      memory    swap          io     system cpu
> r  b  w   swpd   free   buff  cache  si  so    bi    bo   in    cs  us sy  id
>...
> 0  4  7      0 592748  62792 177188   0   0  3780   188  762   935   0 2  98
>  
>
> 
>
>[the secondary]
>
>Device:            tps   Blk_read/s   Blk_wrtn/s   Blk_read   Blk_wrtn
>dev8-0          336.00         0.00      6496.00          0       6496
>
>   procs                      memory    swap          io     system cpu
> r  b  w   swpd   free   buff  cache  si  so    bi    bo   in    cs  us sy  id
>...
> 0  0  0      0 983316   4876  18260   0   0     0  3668  929  1332   0 5  95
>
>  
>
>  
>
>>>>>0: cs:SyncingAll st:Secondary/Primary ns:0 nr:1435652 dw:1435652 dr:0 pe:0 ua:15
>>>>>        [=====>..............] sync'ed: 26.8% (3668/5004)M
>>>>>        finish: 0:47:57h speed: 1,319 (1,313) K/sec
>>>>>1: cs:SyncingAll st:Secondary/Primary ns:0 nr:1367248 dw:1367248 dr:0 pe:0 ua:15
>>>>>        [=====>..............] sync'ed: 26.7% (3672/5004)M
>>>>>        finish: 0:47:40h speed: 1,336 (1,310) K/sec
>>>>>          
>>>>>
>
>  
>
Could you give the config of each server (cpu, memory, disk adapter, 
nic...) the output on each node of:

cat /proc/version
cat /proc/drbd
cat /etc/drbd.conf
cat /var/lib/drbd/drbd.conf.parsed
/sbin/drbdsetup /dev/nb0 show
/sbin/drbdsetup /dev/nb1 show

Start only one synchro:

/sbin/drbdsetup /dev/nbX replicate

During the synchro check the output of:

while true ; do cat /proc/drbd; sleep 1; done
ps -ef
iostat -k 1
vmstat 1

I have 3 clusters running Redhat 9, heartbeat and drbd (2x100Mb/s NICs 
bonding) without problem if you want I could send you a configs.

>bye & thanks for any help,
>
>-christian-
>
>P.S.: Changing the nicelevel below zero is not a good idea as the system
>      becomes quite laggy without any visible speed gain.
>  
>
Best regards.

Francis




More information about the drbd-user mailing list