CUPS 1.2.2 - cups-polld exited on AIX client

angelb angelb at bugarin.us
Wed Aug 2 09:00:47 PDT 2006


Hello all;

Here's another client, an AIX, which I recently installed CUPS 1.2.2.
And it too is exhibiting the same issue where the cups-polld gets SIGTERM from the parent. As you can see from the truss output, the
two cups-polld PIDs are sent a SIGTERM or signal 15 from the parent
cupsd:

1638478:        2232461: kwrite(1, " d   [ 0 2 / A u g / 2 0".., 78) = 78
1732680:        2240613: sendto(4, 0x2FF220E0, 161, 0, 0x2FF210D0, 16)
1638478:        2232461: kread(3, " D E B U G :  \0\0\0\0\0".., 1023) = 71732680
:       2240613: kwrite(2, " [ c u p s - p o l l d  ".., 25)
 = 25
1638478:        2232461: kwrite(1, " E   [ 0 2 / A u g / 2 0".., 86) = 86

1638478:        2232461: close(3)                               = 0
1732680:        2240613: kwrite(2, " 1 1 6 8", 4)               Err#32 EPIPE
1732680:            Received signal #13, SIGPIPE [ignored]
1638478:        2232461: close(3)                               Err#9  EBADF
1732680:        2240613: kwrite(2, " D E B U G 2 :  ", 8)       Err#32 EPIPE
1732680:            Received signal #13, SIGPIPE [ignored]
1638478:        2232461: kwrite(1, " d   [ 0 2 / A u g / 2 0".., 78) = 78
1732680:        2240613: sendto(4, 0x2FF220E0, 161, 0, 0x2FF210D0, 16)
1638478:        2232461: kill(1732680, 15)                      = 0
1732680:            Received signal #15, SIGTERM [default]
1638478:        2232461: kill(466996, 15)                       = 0
1638478:        2232461: _select(65534, 0x20007208, 0x20009218, 0x00000000, 0x2F
F1E8D8) = 1
1732680:        *** process killed ***
, SIGCHLD [caught]
1638478:        2232461: ksetcontext_sigreturn(0x2FF1E480, 0x00000000, 0x2FF1E48
0, 0x2FF3B000, 0x100018FC, 0x0000D0B2, 0x1FBC5400, 0x2FF39F28)


So now, I have 3 clients, 2-AIX(5.2) 1-Solaris(10), all experiencing
the same issues. BTW, these clients used to run CUPS 1.1.23 without
a problem.

A bit of background and shameless disclosure, the clients and servers
are running a TWW-packaged version of CUPS 1.2.2. TWW packages CUPS
for our AIX, Solaris and RedHat systems so that we have consistent
installs and among other things. And afaik, they don't modify the CUPS source but I know they report bugs as they find them. We are currently
running a TWW-packaged version of CUPS 1.1.23 for our development and
production systems all of which are running fine in the same config,
with a bit of modification, as I have been testing in CUPS 1.2.2. The
only drawback with 1.1.23 that I know is its bug for consuming CPU.

So, with that in mind, how do we debug cupsd/cups-polld to see why
cupsd sends a SIGTERM. Obviously, the truss output does not provide
enough granual details to show why cupsd decided to send SIGTERM to its
child processes.


Just a follow-up:

The following clients running two cups-polld daemons are still alive
since yesterday. These daemons were started outside of cupsd to see if
they live long enough and it appears they do. This could be my work-
around until such time the cupsd or related processes, IMHO is
resolved.

sapti02:./cups> ps -ef|grep -i cups
    root 372938      1   0 11:37:16      -  0:11 /opt/TWWfsw/cups122/lib/daemon/cups-polld occam503 631 2700 631
    root 409688      1   0 11:37:16      - 21:07 /opt/TWWfsw/cups122/sbin/cupsd -c /etc/opt/TWWfsw/cups122/cupsd.conf
    root 753898      1   0 11:37:16      -  0:11 /opt/TWWfsw/cups122/lib/daemon/cups-polld occam002 631 2700 631

sapts02f# ps -ef|grep -i cups
    root 20154     1   3 17:07:46 ?          24:46 /opt/TWWfsw/cups122/sbin/cupsd -c /etc/opt/TWWfsw/cups122/cupsd.conf
    root 20155     1   0 17:07:46 ?           0:10 /opt/TWWfsw/cups122/lib/daemon/cups-polld occam503 631 2700 631
    root 20156     1   0 17:07:46 ?           0:10 /opt/TWWfsw/cups122/lib/daemon/cups-polld occam002 631 2700 631

As alsways, comments or suggestions are appreciated.

Thanks,
Angel




More information about the cups mailing list