[SunHELP] random login issue and process problems

[Jarrett Carver]

What you need to do is to crash the server when this hang occurs. (Stop-A, 
then `sync` at OK prompt.) capture the crash for use in figuring out what 
caused the problem.

If resources are not a problem, you could also turn on process accounting. 
You could then use these records after the hang to try and piece together 
what process may have run amuck.

Finding out what was running helped narrow it down tremendously last time I 
had that situation occur. (we found a user attempting to compile something 
on an extremely loaded server...damn users).



Jarrett Carver           http://www.geocities.com/solarboyz1
solarboyz1 at hotmail.com   Unix/NT Systems Administrator
___________________________________________________________
"When people sigh 'Life is hard', I am always tempted to
ask 'compared to what?'".
               - Sydney Harris
____________________________________________________________




>From: "Markham, Richard" <RLMarkham at hafeleamericas.com>
>Reply-To: The SunHELP List <sunhelp at sunhelp.org>
>To: "'sunhelp at sunhelp.org'" <sunhelp at sunhelp.org>
>Subject: [SunHELP] random login issue and process problems
>Date: Tue, 20 Jan 2004 20:27:52 -0500
>
>I wasn't able to login today using console, ssh or telnet on my headless
>V880.
>The login method used will authenticate but the session hangs and does not
>give
>a shell prompt.  At this same time our users lost their WEB (FORMS)
>connections
>to the Oracle E-Business suite.  The server would respond to ping, that's
>about
>it.  After having to hard reset the server I check /var/adm/messages which
>had
>no errors in it.  The server has 8gb RAM, 4CPU's an performs exceptionally
>well.
>This problem exhibits itself out of the blue.  I have allocated a 6gb slice
>for
>swap.  I have a hard time believing that this issue is resource related
>unless
>there are default maximums that I'm unaware of.  I report daily on resource
>usage
>and we have lots free. The same OS account spawns all processes.
>
>My /etc/system shared mem and semaphores are as follows:
>set shmsys:shminfo_shmmax=8388608
>set shmsys:shminfo_shmmin=1
>set shmsys:shminfo_shmseg=10
>set semsys:seminfo_semopm=100
>set semsys:seminfo_semmns=0x13d
>set semsys:seminfo_semmni=0xc
>set semsys:seminfo_semmsl=0x81
>set semsys:seminfo_semmap=0x13d
>set semsys:seminfo_semmnu=0x13d
>
>
>17 days ago I had this exact same problem.  This ~exact~ same problem has
>occurred
>15 days on my SF4800 database server.  We recently upgraded the servers 
>from
>two
>E420R's.  The only thing that I can think of that is significantly 
>different
>is
>that the Recommended patch cluster is 6 months newer (DEC 04,2004) over the
>configuration we have had in place on lesser servers for the past two 
>years.
>I haven't had to set any user session or number of process settings over 
>the
>past
>three years, but if anyone could advise anything worth trying I'm game.
>
>Thanks in advance for anyone's input.  I currently screening the January 
>19,
>04
>patch cluster to be applied this weekend.
>
>Solaris 8 DEC04 '04 Recommended Cluster
>Oracle E-Business Suite.
>_______________________________________________
>SunHELP maillist  -  SunHELP at sunhelp.org
>http://www.sunhelp.org/mailman/listinfo/sunhelp

_________________________________________________________________
Rethink your business approach for the new year with the helpful tips here. 
http://special.msn.com/bcentral/prep04.armx