Re: [rdiff-backup-users] rdiff-backup completes without errors, but then have a corrupt archive...

[Andrew Ferguson]

On Mar 31, 2009, at 2:36 PM, address@hidden wrote:

> > > Are the exit codes the same, and available to cygwin event though
> > > it's a windows binary? (I do see the output is the same unix style
> > > output (with a unix vs dos line return).
>
> > The exit codes should be the same, but, truthfully, the state of  
> > rdiff-
> > backup's exit codes are not very good. I don't think they're
> > documented anywhere (someone correct me if I'm wrong). There's even a
> > bug open in Ubuntu #128244 which basically amounts to "rdiff-backup
> > has unhelpful error codes".
>
> > That said, I would be more than happy to implement useful error  
> > codes.
> > I just don't know what would be useful to system administrators. :-)
>
> > In that bug,
> > https://bugs.launchpad.net/ubuntu/+source/rdiff-backup/+bug/128244
> >  , I proposed:
> > 0 = no error occurred
> > 1 = unrecoverrable error
> > 2 = another rdiff-backup instance running
> > 3 = recoverrable error (link down, interrupted, etc.)
>
> > but no one responded whether those would be good or not, so nothing
> > became of it...
>
> These seem like great exit codes.
> I'm not sure if no 3 covers file errors like files being busy etc...
> I'd guess that even if it did, I'd like a code that is "basically
> everything went smoothly, but some files were locked, or otherwise
> unavailable."


Yes, I completely mangled the words "recoverable" and "unrecoverable"  
as used by: http://www.nongnu.org/rdiff-backup/error_policy.html

what I meant was:

1 = errors which will require sysadmin intervention: eg, CRC failure,  
"assert index out of order", file not found, directory unreadable,  
etc. (ie, unrecoverable by a script)

3 = errors which should go away on the next run of the script  
(assuming the link comes back up, etc.) ... I suppose 2 = another  
backup instance running is a special case of these...


Clearly, I should find new terms to describe those classes of errors,  
since they are all "unrecoverable" as used by that document. I feel  
that all "recoverable errors" as used by that document should be error  
code 0, since rdiff-backup has handled them. If you care about whether  
they occurred or not, then that is what the logging is for.

I don't want *exit codes* to become a terse version of the extensive  
error reporting which rdiff-backup already does. I think exit codes  
should indicate why rdiff-backup exited abnormally.


Andrew