Hans-Dieter.Doll
2008-05-24 21:36:03 UTC
Hi all,
are amdump exit codes somehow classified?
I mean, is there any way to check whether an exit code should be
considered as a warning or a real error (or may be ignored)?
Background: my script calling amdump publishes the result on a
web page, in a way that my service department is alarmed when errors
appear. I'm currently using amanda version 2.5.2p1. Previous versions
seem to return an exit code !=3D 0 on real errors only (something like
no or wrong tape in the tape drive, or whatever).
Since 2.5.2p1 I frequently get exit code 2 (e.g.) because
? gtar: ./hd2/Mail/DrB/Customer/Bosch: file changed as we read it
which is actually harmless, and I would like to suppress any message
if this happens.
With previous amanda versions I did not always get exit codes if errors
appeared someone should care about. Thus my scripts check the amdump out=
put
(i.e. the email report) for strings possibly indicating an error
(and e.g. only warn against strange dump results).
But this approach is for sure not very reliable, and I really like that
newer versions indicate errors by means of exit codes relativly reliable=
.
But I require some kind of classification (otherwise no one cares about
real errors).
How do other people cope with this subject?
Any ideas appreciated...
--hd^2
are amdump exit codes somehow classified?
I mean, is there any way to check whether an exit code should be
considered as a warning or a real error (or may be ignored)?
Background: my script calling amdump publishes the result on a
web page, in a way that my service department is alarmed when errors
appear. I'm currently using amanda version 2.5.2p1. Previous versions
seem to return an exit code !=3D 0 on real errors only (something like
no or wrong tape in the tape drive, or whatever).
Since 2.5.2p1 I frequently get exit code 2 (e.g.) because
? gtar: ./hd2/Mail/DrB/Customer/Bosch: file changed as we read it
which is actually harmless, and I would like to suppress any message
if this happens.
With previous amanda versions I did not always get exit codes if errors
appeared someone should care about. Thus my scripts check the amdump out=
put
(i.e. the email report) for strings possibly indicating an error
(and e.g. only warn against strange dump results).
But this approach is for sure not very reliable, and I really like that
newer versions indicate errors by means of exit codes relativly reliable=
.
But I require some kind of classification (otherwise no one cares about
real errors).
How do other people cope with this subject?
Any ideas appreciated...
--hd^2