Re: Oracle10g running on Solaris Containers

From: Tanel Poder <tanel_at_poderc.com>
Date: Sun, 1 Nov 2009 01:00:53 +0800
Message-ID: <4602f23c0910311000s47f4fef0ybe753d5456cb46fd_at_mail.gmail.com>

If there are crashes, there usually are crashdumps (written by Oracle) or coredumps (written by OS signal handlers).

So, you could check if you see any files in following directories:

1. /var/adm/crash
2. /var/cores
3. whatever is specified in Oracle's core_dump_dest param

Also run ulimit -a as your Oracle user to make sure that corefile sizes haven't been set to 0:

tanel_at_solaris01:~$ ulimit -a

core file size          (blocks, -c) *unlimited*
data seg size           (kbytes, -d) unlimited
file size               (blocks, -f) unlimited
open files                      (-n) 256
pipe size            (512 bytes, -p) 10
stack size              (kbytes, -s) 10240
cpu time               (seconds, -t) unlimited
max user processes              (-u) 29995
virtual memory          (kbytes, -v) unlimited

If you find an OS-generated corefile, you can run pstack on it to see in which functions the crash happened and put these functions into metalink or send to support.

Without a corefile (or any logfile entries returning the signal/error code why a process was forced to exit) its guesswork...

-- 
Tanel Poder
http://blog.tanelpoder.com


On Wed, Oct 28, 2009 at 10:54 PM, <Mayen.Shah_at_lazard.com> wrote:



>

> Hello everyone,

>

> We recently migrated few database (10.2.0.3 and 10.2.0.4) from Solaris 9

> running on E25k to Solaris 10 running on T5120 running multiple zones.

> Entire directory structure was moved from old machine to new machine.

> Oracle relinking was successful.

>

> Since then, database are crashing frequently. mainly process (pmon or smon)

> would simply die. Unix sas could not find any thing. I could not find any

> thing either.

> One thing I noticed, on old machine I had only one dbwriter dbw0 but on new

> machine I have upto six or seven dbwrites (dbw0, dbw1,....). No parameter

> change have been made. I am not sure why this change in number of dbwrites

> and if it has any relevance.

>

> I have forwarded metalink note 317257.1 to UNIX sas.

>

> Does any one has been running Oracle10g on Solaris containers? Any

> problems/issues?

>

> Any help/pointers are greatly appreciated.

>

> Thank you

>

> Mayen

>




-- 
Tanel Poder
http://blog.tanelpoder.com

--
http://www.freelists.org/webpage/oracle-l

Received on Sat Oct 31 2009 - 12:00:53 CDT