non-killable program - kernel problem?

non-killable program - kernel problem?

[Erik Steffl]

  cleanup (part of postfix) is eating up most of CPU cycles and I cannot
kill it (-9 is ignored) - I guess that means that the problem is
somewhere in kernel (system call). Is it true? Is there anything I can
do to kill it? to find out what the problem is?

  I attached gdb to it but now gdb is stuck as well... I didn't see
anything suspicious in system log...

  kernel 2.4.18, I didn't upgrade it for a long time and this cleanup
problem only occured recently (last few weeks, twice).

  TIA

	erik

Re: non-killable program - kernel problem?

[Erik Steffl]

Erik Steffl wrote:
> 
>   cleanup (part of postfix) is eating up most of CPU cycles and I cannot
> kill it (-9 is ignored) - I guess that means that the problem is
> somewhere in kernel (system call). Is it true? Is there anything I can
> do to kill it? to find out what the problem is?
> 
>   I attached gdb to it but now gdb is stuck as well... I didn't see
> anything suspicious in system log...
> 
>   kernel 2.4.18, I didn't upgrade it for a long time and this cleanup
> problem only occured recently (last few weeks, twice).
> 
>   TIA

  now this is really strange - after FEW DAYS of load 99% and not being
able to interrupt cleanup (I attached gdb to it and hit ctrl-c) it was
finally interrupted and now it's in:

(gdb) where
#0  0x401f82e4 in open () from /lib/libc.so.6
#1  0x4004354c in rewrite_clnt_stream () from
/usr/lib/libpostfix-global.so.1
#2  0x40038151 in mail_stream_file () from
/usr/lib/libpostfix-global.so.1
#3  0x0804be05 in dict_changed ()
#4  0x08049a65 in dict_changed ()
#5  0x40027b85 in _init () from /usr/lib/libpostfix-master.so.1
#6  0x40027ce1 in _init () from /usr/lib/libpostfix-master.so.1
#7  0x40051e05 in event_loop () from /usr/lib/libpostfix-util.so.1
#8  0x400286ce in single_server_main () from
/usr/lib/libpostfix-master.so.1
#9  0x08049c4e in dict_changed ()
#10 0x4014e14f in __libc_start_main () from /lib/libc.so.6
(gdb) 

  I guess that since it's still in open function of libc it was stuck in
open system call?

  any ideas about what's going on?

	erik

Re: non-killable program - kernel problem?

[Pavel Machek]

Hi!

>   cleanup (part of postfix) is eating up most of CPU cycles and I cannot
> kill it (-9 is ignored) - I guess that means that the problem is
> somewhere in kernel (system call). Is it true? Is there anything I can
> do to kill it? to find out what the problem is?

Yep, that's a kernel bug. Power switch should kill it safely
;-). Magic keys may be able to show where it loops.
									Pavel

-- 
(about SSSCA) "I don't say this lightly.  However, I really think that the U.S.
no longer is classifiable as a democracy, but rather as a plutocracy." --hpa