Amanda-Users

Re: Problem after years of backing up successful: "data timeout"

2006-12-15 03:56:42
Subject: Re: Problem after years of backing up successful: "data timeout"
From: Paul Bijnens <paul.bijnens AT xplanation DOT com>
To: Jobst Schmalenbach <jobst AT barrett.com DOT au>
Date: Fri, 15 Dec 2006 09:47:34 +0100
Jobst Schmalenbach schreef:
All,

I am replying to my own email cause I fixed it ...
however, I am not sure whether this is a bug!

I tried a lot of things to find the problem ... until I did
a "mount" to display the mounted devices. I just wanted
to see whether a USB HD (which I use for the ODD backup of the
system and this thing is BIG) is mounted.

I saw that a WINDOWS workstation which I had mounted
(to check the size of a partition and the used filespace)
was still mounted ... (the amanda host is a samba host too).

Further I saw lots of thesew messages in the logs
  ...
  smb_add_request: request [3d6a9e00, mid=1646] timed out
  ...

I umounted the windows workstation partition and everything is fine now ...


What I find strange:

 * If I have a drive mounted from a UNIX box and the
   the Unix workstation is now turned off it doesnt matter,
   this does NOT affect amanda.

 * The windows workstation in question is NOT part
   of the disklist (I do not backup any windows boxen)
   and I do not have any windows based servers.

 * all machines backed up by Amanda are UNIX based.


So how can a windows workstation that has one of its partition mounted and is now turned of affect Amanda although
the workstation is not part of the daily backup routine????


If I understand correctly, the amanda client (Unix) had a remote
mounted filesytem form an Windows workstation (mounted with smbmount).
That Windows workstation was turned off.

Seems indeed a problem with smbfs, see this thread:

http://lists.samba.org/archive/samba/2004-August/090271.html

Or these questions without solution:

http://lists.samba.org/archive/smb-clients/2003-February/000257.html
http://lists.samba.org/archive/samba/2004-October/094082.html

That last post asks for the smb equivalent for the NFS options
"-o hard,intr", while he actually means "-o soft", I believe.
A NFS "hard" mount will wait indefinetly for a nonresponsive remote
system. In contrast: a syscall will return an error for a "soft" mounted
unresponsive remote system after some timeout (about a minute or so).

To be complete, for NFS, you usually DO want a "hard" mount, because
most programs are not prepared to handle error intelligently for
failed syscalls on soft mounted filesystems.  And together with "hard",
you usually also want "intr", so that you at least interrupt the
program when you type e.g. accidently "ls" on such a place.

Seems like smb has no such option, and defaults to hang.
   ;-) seems reasonable, as most MS users are already familiar with
   hangs, more than with a well formulated error message :-)

Using gnutar with Amanda, on such "hard" mounted unresponsive remote
filesystems, hangs too, even if you did not add that filesystem
to the DLE's.  In order to find out if the directory is on a remote
filesystem (and to stop descending into it), gnutar needs to do
a stat(), and that will hang.  It's only after the stat-call returns
that gnutar would know this is a remote filesystem.


--
Paul Bijnens, Xplanation                            Tel  +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUM    Fax  +32 16 397.512
http://www.xplanation.com/          email:  Paul.Bijnens AT xplanation DOT com
***********************************************************************
* I think I've got the hang of it now:  exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit,  ZZ, :q, :q!,  M-Z, ^X^C,  logoff, logout, close, bye,  /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt,  abort,  hangup, *
* PF4, F20, ^X^X, :D::D, KJOB, F14-f-e, F8-e,  kill -1 $$,  shutdown, *
* kill -9 1,  Alt-F4,  Ctrl-Alt-Del,  AltGr-NumLock,  Stop-A,  ...    *
* ...  "Are you sure?"  ...   YES   ...   Phew ...   I'm out          *
***********************************************************************

<Prev in Thread] Current Thread [Next in Thread>