HP3000-L Archives

January 2003, Week 5

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Reply To:
Date:
Thu, 30 Jan 2003 12:32:02 -0600
Content-Type:
text/plain
Parts/Attachments:
text/plain (66 lines)
 JIM McINTOSH wrote:
<SNIP>
 >In the meantime, APACHE still services other (i.e., non CGI requests)
 >just fine.
</SNIP>
Jim,

I don't know about the specific HP3000 situation (I do know the 3000 but
haven't used the shell that often).

I have, however, seen similar behaviour on a 9000 where other processes
consumed all the system resources so there were none left for apache. It
is logical that the 'normal' serving of webpages continues because that
is done within the running apache processes. To fullfill the cgi
requests apache needs to spawn a child process from which the output is
served to the user by one of the mentioned serverprocesses, this is also
why apache is perfectly capable of returning an error message both to
the logs and the browser, even when it cannot spawn anymore children.

Maybe you need to look into specific shell settings to determine how
many processes a single user is allowed to start and/or the maximum
total of processes sh.hpbin will allow.
It may even be (but I'd have to look in the 3K manuals for that) that a
new instance of sh.hpbin is launched for each (child-) process and that
you run into your machine's mpe job limit, but that's a wild guess.

<SNIP>
> We first attempted to stop APACHE using the command:  xeq sh.hpbin.sys "-c
> 'kill `cat logs/httpd.pid`'"
> The job did not stop.  What is in the error_log is:
>
> [Wed Jan 29 09:37:49 2003] [error] child process 6881764 still did not exit,
> sen
> ding a SIGKILL
> [Wed Jan 29 09:37:49 2003] [error] child process 7144436 still did not exit,
> sen
> ding a SIGKILL
> [Wed Jan 29 09:37:49 2003] [error] child process 10880224 still did not
> exit, se
> nding a SIGKILL
> [Wed Jan 29 09:37:50 2003] [notice] caught SIGTERM, shutting down
>
> When we attempt to perform an abortjob on the job, there is no response.
>


The shutdown errors you got do not look any weirder to me than a forced

shutdown of an unwilling apache on HPUX.


HTH

Jur

--
======================================================
              Remove whatever you like.
    The only spam I can stand requires a can-opener.
======================================================



* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2