Re: [Cooker] CUPS update on 8.2 -- b0rk3n down the system!

2003-01-18 Thread Claudio
Hi again!
Thanks for your suggestion: the problem does not appear if I *manually*
stop cups before executing the update! Maybe the rpm-script does not shut
down cups properly if a client is sending a job to the server at that
moment... Anyway it's not a packaging problem I suppose :o)

  Claudio


 On Wed Jan 15, 2003 at 11:12:53AM +0100, Claudio wrote:
[...]
 [...]
 I [15/Jan/2003:10:16:07 +0100] Configured for up to 100 clients. I
 [15/Jan/2003:10:16:07 +0100] Allowing up to 10 client connections per
 host. I [15/Jan/2003:10:16:07 +0100] LoadPPDs: Read
 /etc/cups/ppds.dat, 1021 PPDs...
 I [15/Jan/2003:10:16:08 +0100] LoadPPDs: Wrote /etc/cups/ppds.dat,
 1021 PPDs...
 E [15/Jan/2003:10:16:14 +0100] StartListening: Unable to bind socket -
 Cannot assign requested address.

 This looks to me like cups didn't stop clean before.  If it's unable to
 bind to the socket, that means something has got a hold of the socket
 already, meaning that during the upgrade when cups was supposed to be
 stopped, it didn't really stop.  IIRC, the initscript tries to kill the
 cupsd based on the pid in the /var/run/cupsd.pid (or whatever) file.  If
 that file is removed, you effectively have a cupsd that is not
 controllable via the initscript (initscript thinks cupsd is already
 stopped because the pid file is no longer present).

 Downgrading to 8.2 original packages, server comes back working
 fine!

 Try what I suggested above.  I'm quite certain there should be nothing
 wrong with these cups packages as I tested them here and was able to run
 them and print through them.  IIRC, I didn't have cups previously
 installed/configured, so there might be some strangeness with the
 upgrade that I missed, but I don't think so.  The bind socket error
 tells me something different.



-- 







Re: [Cooker] CUPS update on 8.2 -- b0rk3n down the system!

2003-01-15 Thread Vincent Danen
On Wed Jan 15, 2003 at 11:12:53AM +0100, Claudio wrote:

 Hi all!
 I send this to the list, since I don't know who is responsible for this
 kind of problems  :-)
 I updates CUPS as you can see from /var/log/messages:
 
 Jan 15 10:15:49 sertov2 perl: [RPM] libcups1-1.1.18-1.1mdk installed
 Jan 15 10:15:49 sertov2 perl: [RPM] libcups1-1.1.14-2mdk removed
 Jan 15 10:15:49 sertov2 perl: [RPM] libcups1-devel-1.1.18-1.1mdk installed
 Jan 15 10:15:49 sertov2 perl: [RPM] libcups1-devel-1.1.14-2mdk removed
 Jan 15 10:15:50 sertov2 perl: [RPM] cups-common-1.1.18-1.1mdk installed
 Jan 15 10:15:51 sertov2 perl: [RPM] cups-common-1.1.14-2mdk removed
 Jan 15 10:15:55 sertov2 perl: [RPM] cups-1.1.18-1.1mdk installed
 Jan 15 10:16:04 sertov2 cups: cupsd shutdown succeeded
 Jan 15 10:16:08 sertov2 kernel: lp0: compatibility mode
 Jan 15 10:16:08 sertov2 last message repeated 2 times
 Jan 15 10:16:14 sertov2 cupsd: cupsd: Child exited with status 99!
 Jan 15 10:16:15 sertov2 cups: cupsd startup succeeded
 
 [root@sertov2 root]# /etc/rc.d/init.d/cups status
 cupsd is stopped

Could be that during the upgrade of the various cups components it
shuts the cups server off instead of trying to restart it.

 Trying to restart it, implies that cups reports a failed when shutting
 down and ok when starting up, but it says stopped when I ask the
 status.

Ok...  if you restart a server that is stopped, you will get a failed
on the stop...  it's already stopped so it can't stop it.  That's why
you get your failed.  It does start up however, right?  You said it
gives ok when it starts.

The status thing I'm not sure of...  you did the status after you
restarted?  Try doing a service cups stop, then do ps ax|grep cups
and make *sure* it is stopped.  Then do a service cups start and then
ps ax|grep cups and make sure cups is running.  Finally, do your
service cups status and see if it still says cups is stopped.  It
might not be writing the pid file properly, so the status might be
misleading.

Of course, it is entirely possible that cups isn't starting, but at
least by doing the above, you will know for sure.

 I have both cups and cups-lpd started at boot time on my printerserver,
 and samba+netatalk are present too.
 The only thing I could find in /var/log/cups/error_log is
 
[...]
 I [15/Jan/2003:10:16:07 +0100] Configured for up to 100 clients.
 I [15/Jan/2003:10:16:07 +0100] Allowing up to 10 client connections per host.
 I [15/Jan/2003:10:16:07 +0100] LoadPPDs: Read /etc/cups/ppds.dat, 1021
 PPDs...
 I [15/Jan/2003:10:16:08 +0100] LoadPPDs: Wrote /etc/cups/ppds.dat, 1021
 PPDs...
 E [15/Jan/2003:10:16:14 +0100] StartListening: Unable to bind socket -
 Cannot assign requested address.

This looks to me like cups didn't stop clean before.  If it's unable to
bind to the socket, that means something has got a hold of the socket
already, meaning that during the upgrade when cups was supposed to be
stopped, it didn't really stop.  IIRC, the initscript tries to kill the
cupsd based on the pid in the /var/run/cupsd.pid (or whatever) file. 
If that file is removed, you effectively have a cupsd that is not
controllable via the initscript (initscript thinks cupsd is already
stopped because the pid file is no longer present).

If you do have a zombie cupsd running that the initscript cannot
stop, you can try to kill -9 it.  This should not normally happen, but
it's possible something/someone was trying to print which may have
prevented cupsd from stopping when you wanted to stop it.  Of course, I
could have no clue what I'm talking about since I don't know much about
cups other than that it has always just worked.  =)  But this is what I
would look for.

 Downgrading to 8.2 original packages, server comes back working fine!

Try what I suggested above.  I'm quite certain there should be nothing
wrong with these cups packages as I tested them here and was able to
run them and print through them.  IIRC, I didn't have cups previously
installed/configured, so there might be some strangeness with the
upgrade that I missed, but I don't think so.  The bind socket error
tells me something different.

-- 
MandrakeSoft Security; http://www.mandrakesecure.net/
lynx -source http://linsec.ca/vdanen.asc | gpg --import
{FE6F2AFD : 88D8 0D23 8D4B 3407 5BD7  66F9 2043 D0E5 FE6F 2AFD}


msg86197/pgp0.pgp
Description: PGP signature