>> i searched the list archives but couldn't find anything appropriate
>> regarding this problem:
>>
>> we deploy squid on IBM RS/6000 machines running AIX 4.3.1.0
>> as well as 4.3.2.0, and we're experiencing very frequent core
>> dumps and erratic behaviour of the application -- squid usually
>> dies spontaneously after about an hour of operation. unfortunately
>> there's not much information in the logs to be found (with one exception,
>> see below):
>>
>> > /squid/bin/RunCache[22]: 15358 Illegal instruction(coredump)
>
> In my experience this usually points to memory allocation problems,
>independently of what application you're running. Since these days
>compilers are usually good enough not to generate invalid code, usually
>it means the processor was sent somewhere strange due to a corrupted
>function pointer.
>
> Have you tried recompiling and relinking Squid with a different set
>of malloc routines than what you're using now? If that doesn't do it,
>you may want to turn on some memory allocation debugging code and see
>if it detects anything helpful.
> -- Clifton
Ok, I compiled in the dlmalloc code. Squid still dumped core about at
the same frequency as before.
I got GDB (instead of dbx) , compiled it and tried to get it to read
the core file, but it can't. It says:
"/var/squidcache/core" is not a core dump: File format not recognized
Would GNUmalloc help me possibly?
I'll try to turn on malloc debugging. See if this helps.
Regards
Chris
Received on Thu Jul 22 1999 - 03:44:56 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:28 MST