Re: [rdiff-backup-users] ridff-backup 'hangs' on certain file - update

[Maarten Bezemer]

Hi,

Could you try making a new repo containing only the directory that 
includes the 'broken' file, and see if that fails too?
(How large is that file anyway?)

If that setup also fails, create a tar or use rsync to copy it to the 
backup server, and recreate the source tree there, to test if it works 
when running rdiff-backup locally.

I've seen a few cases (not with rdiff-backup but other software) where the 
network hardware somewhere between the hosts was malfunctioning. Smaller 
packets went through fine, larger packets got mangled and subsequently 
dropped. Ssh went fine, scp on a big file hung up.
Maybe setting MTU to a smaller value can be a workaround for your problem.

On the other hand, your strace results (I/O error) are not what's to be 
expected... did you try rebooting? Did you try memtest86+ to test your RAM?


HTH,

Maarten


On Sun, 16 May 2010, Danilo Godec wrote:

> I removed 'rdiff-backup-data' for this host and tried the 'rdiff-backup'
> again - it still stops and 'hangs' on the same file.
>
> I interrupted the 'rdiff-backup' process (CTRL-C), removed all backed-up
> files for this host and 'rsync'-ed it - without any problems.
>
> Regards, Danilo
>
>
> On 15.5.2010 12:42, Danilo Godec wrote:
> > Hi,
> >
> > recently my 'rdiff-backup' developed a weird problem, where it 'hangs'
> > when backing up a certain file on one server. Other 40+ servers are OK,
> > it's just that one and even that is only happening since May 3rd....
> >
> > I can 'cat' the file on the originating server, I can also 'scp' it on
> > the backup server - there is no problem, no error with that. However -
> > when 'rdiff-backup' gets to this file, it just 'hangs' and does nothing.
> >
> > On the backup server I see the file 'rdiff-backup.tmp.22397' which seems
> > the be a partially transferred original file (524288 bytes vs. 785592
> > bytes of the original file).
> >
> > If I 'strace' the 'python' process on the backup server, I get this:
> >
> >
> > > # strace -p 7343
> > > Process 7343 attached - interrupt to quit
> > > read(5, ^C <unfinished ...>
> > > Process 7343 detached
> > If I strace the 'ssh' process', I get this:
> >
> >
> > > # strace -p 7344
> > > Process 7344 attached - interrupt to quit
> > > select(7, [3 4], [], NULL, NULL^C <unfinished ...>
> > > Process 7344 detached
> > And that's all, there is nothing else going on even if I leave 'strace'
> > open for 30 minutes...
> >
> > And if I 'strace' the 'python' process on the originating server, I get
> > this:
> >
> >
> > > # strace -p 20518
> > > Process 20518 attached - interrupt to quit
> > > read(3,  <unfinished ...>
> > > strace: ptrace(PTRACE_CONT,1,133): Input/output error
> > > Process 20518 detached
> > After that the process state in 'ps' changes from 'Ss' to 'Ts'
> > (stopped). I can change it back to 'Ss' with 'kill -CONT', but it still
> > doesn't do anything.
> >
> > The weird thing is that it ALWAYS happens on the same file, but there is
> > seemingly nothing wrong with that particular file...
> >
> > Any ideas? What else is there to try and get more clues?
> >
> >    D.
> >
> > PS: OS of the backup server is OpenSuSE 11.1 (32 bit), OS of the
> > 'backed-up' server is CentOS release 5.2 (64 bit). Rdiff-backup version
> > on both is 1.2.8. I also tried removing 'rdiff-backup-data' to start all
> > over, but it didn't help.
> >
> >
> > _______________________________________________
> > rdiff-backup-users mailing list at address@hidden
> > http://lists.nongnu.org/mailman/listinfo/rdiff-backup-users
> > Wiki URL: http://rdiff-backup.solutionsfirst.com.au/index.php/RdiffBackupWiki
>
>
> _______________________________________________
> rdiff-backup-users mailing list at address@hidden
> http://lists.nongnu.org/mailman/listinfo/rdiff-backup-users
> Wiki URL: http://rdiff-backup.solutionsfirst.com.au/index.php/RdiffBackupWiki