Tux-6.5 (no patches) BBL crashes with SIGBUS error

Question

Hi,
We have been using tuxedo 6.5 on HP-UX 11i for a long time. All of a sudden, in
the last couple of days we wre finding that the BBL process crashes with SIGBUS
error. Nothing is written into the ULOG to indicate what caused the problem. We
tried to follow the BBL process to see what was happening and the output at the
time of the crash is:
06/30/03 - 12:55:12 unlink("/var/tmp/TUXFTFa09561") ...... = 0
06/30/03 - 12:55:12 time(0) .............................. = 1056974112
06/30/03 - 12:55:12 getpid() ............................. = 9561 (1)
06/30/03 - 12:55:12 saccess("/var/tmp/vtda09561", F_OK) .. ERR#2 ENOENT
06/30/03 - 12:55:12 lstat("/var/tmp/vtda09561", 0x68af6c78) ERR#2 ENOENT
06/30/03 - 12:55:12 open("/var/tmp/vtda09561", O_WRONLY|O_CREAT|O_TRUNC, 0666)
= 10
06/30/03 - 12:55:12 write(10, "100202\0\0\0\0@ \00302c7\0\0\0\0".., 61732) = 61732
06/30/03 - 12:55:12 close(10) ............................ = 0
06/30/03 - 12:55:12 msgsnd(197319, 0x400b2148, 54, IPC_NOWAIT|0) = 0
06/30/03 - 12:55:12 Received signal 10, SIGBUS, in user mode, [SIG_DFL], partial
siginfo
06/30/03 - 12:55:12 Siginfo: si_code: I_NONEXIST, faulting address: 0x3b5,
si_errno: 0
06/30/03 - 12:55:12 PC: 0xc0196af3, instruction: 0x495a0008
06/30/03 - 12:55:13 exit(10) [implicit] .................. WIFSIGNALED(SIGBUS)|WCOREDUMP
I have no idea what is causing the crash. We know for sure that no kernel level
changes have been applied in the last 31 days at least. I would appreciate it
very much if someone could point me in the right direction to understand why we
are getting the BBL crashing.

Answer 1

If you were to contact BEA Customer Support with a problem like this, the first
thing they would tell you is to get the latest rolling patch. It is possible
that this is a problem that was resolved in a patch and people do not want to
spend time looking into a problem that was solved by somebody else.
That said, when a system that was running for a long time starts having trouble,
there has to be something that is different. A few ideas: a disk might be full
or your system is low on memory. Does the problem exist after a reboot?, Did
you change the size of your application? Can you restore a save of the state
of your application before you started having the problem? How about the whole
system? Are there any new applications running that will change the usage of
memory or message queues?
I did a quick look at the patchlev file for Tux 6.5 and there were a number of
problems with the BBL fixed over the years. Several have involved crashes. Some
of the problems are quite subtle and you have to be lucky (or unlucky) to hit
them. It is possible you were running "for a long time" and were just lucky to
be avoiding the problem.
What this comes down to is that crashes of the BBL are particularly disturbing
and I recommend getting the latest rolling patch. If the problem persists, I
suggest filing a case with Customer Support.
Regards,
/RC Bryan
"Love Kisslay" <[email protected]> wrote:
>
Hi,
We have been using tuxedo 6.5 on HP-UX 11i for a long time. All of a
sudden, in
the last couple of days we wre finding that the BBL process crashes with
SIGBUS
error. Nothing is written into the ULOG to indicate what caused the problem.
We
tried to follow the BBL process to see what was happening and the output
at the
time of the crash is:
06/30/03 - 12:55:12 unlink("/var/tmp/TUXFTFa09561") ...... = 0
06/30/03 - 12:55:12 time(0) .............................. = 1056974112
06/30/03 - 12:55:12 getpid() ............................. = 9561 (1)
06/30/03 - 12:55:12 saccess("/var/tmp/vtda09561", F_OK) .. ERR#2 ENOENT
06/30/03 - 12:55:12 lstat("/var/tmp/vtda09561", 0x68af6c78) ERR#2 ENOENT
06/30/03 - 12:55:12 open("/var/tmp/vtda09561", O_WRONLY|O_CREAT|O_TRUNC,
0666)
= 10
06/30/03 - 12:55:12 write(10, "100202\0\0\0\0@ \00302c7\0\0\0\0".., 61732)
= 61732
06/30/03 - 12:55:12 close(10) ............................ = 0
06/30/03 - 12:55:12 msgsnd(197319, 0x400b2148, 54, IPC_NOWAIT|0) = 0
06/30/03 - 12:55:12 Received signal 10, SIGBUS, in user mode, [SIG_DFL],
partial
siginfo
06/30/03 - 12:55:12 Siginfo: si_code: I_NONEXIST, faulting address:
0x3b5,
si_errno: 0
06/30/03 - 12:55:12 PC: 0xc0196af3, instruction: 0x495a0008
06/30/03 - 12:55:13 exit(10) [implicit] .................. WIFSIGNALED(SIGBUS)|WCOREDUMP
I have no idea what is causing the crash. We know for sure that no kernel
level
changes have been applied in the last 31 days at least. I would appreciate
it
very much if someone could point me in the right direction to understand
why we
are getting the BBL crashing.