Erik P. Olsen wrote:
> I am running Bacula 1.38.5 on linux FC5 and this morning a full back-up of my 
> windows 2000 laptop terminated with the following error message:
> 
> 16-Apr 08:59 epowork-fd: epowork.2007-04-16_02.05.00 Fatal error: 
> c:\cygwin\home\kern\bacula\k\src\win32\filed\../../filed/backup.c:500 Network 
> send error to SD. ERR=Input/output error
> 16-Apr 08:59 epowork-fd: epowork.2007-04-16_02.05.00 Error: 
> c:\cygwin\home\kern\bacula\k\src\win32\lib\../../lib/bnet.c:393 Write error 
> sending len to Storage daemon:192.168.1.125:9103: ERR=Input/output error
> 16-Apr 08:59 epo-dir: epowork.2007-04-16_02.05.00 Error: Bacula 1.38.5 
> (18Jan06): 16-Apr-2007 08:59:57
> 
> What is bacula trying to tell me? And where does c:\cygwin\home\kern\.... 
> come from?

Hi,

(Maybe someone experienced on this list can decide if Kern should read 
this, too?)


Solutions:
Can you tell me what kind of network and routing this is going through? 
(hops, NAT, stunnel, VPN, hardware...)

Also, some people have had success with Heartbeat Interval for FD and SD 
in similar situations (this specific error is pretty unique according to 
google, though).

Or try updating the windows network interface card driver, too.

The cygwin thing is from compiling, and says where in the code we are 
going wrong. Don't worry about it.


A possible explanation:
I have been getting the same and similar errors. It seems related to 
networking over slow links (or huge jobs) that go through some 
problematic routing. For me, it happens after a specific amount of time, 
such as 12 hours. Sounds like broken networking to me.

I have been working on it on and off for a month now, and today the 
first 58 hour job finished :). So I think I will have a solution 
sometime soon, but as of yet I have not seen other cases or solutions.

The key seems to be that the DIR->FD connection is cut off during backup 
because of lack of traffic. The FD->SD connection can go on for a long 
time (30 mins or more) after this. The errors suggests a problem between 
FD->SD, of course - but dumping the traffic shows me this is wrong. 
Internally, this can also create weird reports in bacula because the 
director doesn't seem to check with SD that it also finished the job... 
I might file a bug about it when I know more.

Cheers,
Johan

-- 
Johan Ehnberg

Email: [EMAIL PROTECTED]
GSM:   +358503209688
WWW:   http://www.ehnberg.net/johan/

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to