[Dirvish] how to debug problems?

dirvish at rain4us.net dirvish at rain4us.net
Fri Oct 14 07:19:46 PDT 2005


> Hi Guys,
>
> I had a working dirvish setup which collects from 5 machines. One of
> these "vaults" stopped to work properly. In seems that rsync "hangs".
> After some inspection I got the following information:

<snip>

> I thought this was a filesystem problem but the following command works
> perfect (/backup/hede/ is in the same partiton with /backup/_NEW/):
> rsync --dry-run -vvaP  removed:/originals/ /backup/hede/
>

Well, it might STILL be related to the file system.

>
> I think it has something to do with the number of files. What can you
> suggest for further investigation?

Interestingly enough, I too think there is something to the "number of
files on a system" hypothesis.  I use dirvish to backup Courier Imap
Maildirs for about 75 accounts which means QUITE A FEW FILES!  I have
found that my file system has been corrupted SEVERAL times.  Not
necessarily the SAME disk but for SURE the same bank. (I have four
separate banks with two of the banks rotating between two disks.  I have 6
disks in all)

Well, http://www.namesys.com/faq.html#reiserfsspecs debunks THAT theory! 
Still,  something is not right.   Does 'dirvish-runall' kick off ALL the
jobs at once?  The log files seem to indicate some sort of concurrency. 
Is there a way to make it kick off the jobs in sequence?  Is the disk
channel being overloaded?  I wonder if there is a memory problem..hmm..

More suggestions in solving these dirvish 'hangs' would be great.

-- 
Richard


More information about the Dirvish mailing list