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

Francis SOUYRI francis.souyri at apec.fr
Mon Jan 26 14:12:21 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:

>On Mon, Jan 26, 2004 at 11:03:03AM +0100, Francis SOUYRI wrote:
>  
>
>>In the /var/lib/drbd/drbd.conf.parsed I see that you set the 
>>"drbdX:fsckcmd fsck -p -y" if you use a journaling filesystem (reiserfs, 
>>ext3, xfs, jfs) you can set it to "drbdX:fsckcmd /bin/true". 
>>    
>>
>Good idea, but there was no fsck running and journaling filesystems also
>need a fsck every now and then as the journal only prevents problems at
>reboots without clean unmounts but does not care about bit errors and
>corrupt filesystems etc that do (out of experience) seldomly. Especially
>on a Maildir-based mailserver system.
>  
>
Escuse me I do not understand: "that do (out of experience) seldomly".

You normaly need a fsck on a journaling filesystem when you have a 
hardware problem (bit errors, dead blocks), but with drbd, when you have 
a disk hardware problem on the primary drbd device the system "panic" 
(check your drbd config), your operator restart the application on the 
other node where the disk hardware is correct, you only need to recover 
the journal (because the filesystem was not unmouted on the other node).

If you have a disk hardware problem like bit errors, or more certainly 
dead blocks, a fsck can help you to recover a coherent filesystem and 
you can mount it and access some datas, but some datas are dead.

>  
>
>>Could you create another drbd device to do some tests (manuel
>>synchro....) ?
>>    
>>
>Not on this system but maybe I try it on test systems in the near
>future. Sadly I also ran out of testing hardware, esp. I have no
>more of these intel gigabit cards...
>  
>
>>Francis
>>    
>>
>thanks,
>
>-christian-
>
>  
>
Best regards.

Francis




More information about the drbd-user mailing list