Martin Simmons wrote:
>> Well I can't repeat this here with 1.36.3 on FreeBSD. What should happen
is
>> that the director says
>>
>> Fatal error: FD gave bad response to ClientRunBeforeJob command: wanted 2000 OK RunBefore
>> got: 2905 Bad RunBeforeJob command.
Stephan> now we get closer: I modified my ClientRunBeforeJob to exit with 100.
Stephan> And I get exactly what your director says. Here's the output:
Interesting. What is the largest number that behaves like this?
looks like it's 199. But I did not try everything in between. ;-)
200 gives:
21-Sep 16:36 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.36.43 Fatal
error: ClientRunBeforeJob returned non-zero status=268435656.
ERR=Permission denied
201 gives:
21-Sep 16:39 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.39.19 Fatal
error: ClientRunBeforeJob returned non-zero status=268435657. ERR=Exec
format error
202 gives:
21-Sep 16:40 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.39.58 Fatal
error: ClientRunBeforeJob returned non-zero status=268435658. ERR=Bad
address
and 205 is:
21-Sep 16:46 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.46.30 Fatal
error: ClientRunBeforeJob returned non-zero status=268435661. ERR=File
name too long
...
Does returning 5 like you did initially always return in "Backup OK -- with
warnings" with "Non-fatal FD errors: 1"?
as ClientRunAfterJob it does.
as ClientRunBeforeJob returning 5 behaves as expected - the Job fails
with fatal error. The message in that case is rather normal:
21-Sep 16:48 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.48.18 Fatal
error: ClientRunBeforeJob returned non-zero status=268435461. ERR=Child
exited with code 5
Stephan> when I do the same thing as ClientRunAfterJob it does /not/ error out
Stephan> like this [1]. So it looks like there's a difference between
Stephan> ClientRunBeforeJob and ClientRunAfterJob return codes.
Yes, the result of ClientRunAfterJob is probably ignored because it is
pointless making the job fail after it has finished.
hmmm, in my case not. Since I may have stored something that is not
(entirely) useful.
The background of this is that my ClientRunBeforeJob initiates a Oracle
database full export. The actual data is written to a FIFO pipe where
bacula sucks the data from during the backup job. This works pretty well
so far.
Finally I use the ClientRunAfterJob to parse the log file that oracle
creates. Since sometimes the database is not /entirely/ exported. The
dump might not contain everything we would need/want to restore. E.g.
somebody could have locked a single table. Or oracle produced a variety
of other warnings. In that case it is necessary to run the export again.
Thus the backup job should fail and notify us this way.
Stephan
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server.
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users
