[AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Allan Regenbaum DP

We are running a production AOLServer 3.3.1 + ad13 on solaris
We have 2 machines, duplicates of each other.

Machine A suddenly, without error messages stops running scheduled procs !!
Machine B continues to perform flawlessly
The machine is running acs classic 4.2 tcl, 90% webmail which has been
running perfectly well under stress and load for months...


Urgent help/suggestions are appreciated ..

Allan Regenbaum
Digital People



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Barry Books

Did you turn on clustering? I think tasks only run on the master in a
cluster.

Barry Books

-Original Message-
From: Allan Regenbaum DP [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, October 30, 2001 8:03 AM
To: [EMAIL PROTECTED]
Subject: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS


We are running a production AOLServer 3.3.1 + ad13 on solaris
We have 2 machines, duplicates of each other.

Machine A suddenly, without error messages stops running scheduled procs !!
Machine B continues to perform flawlessly
The machine is running acs classic 4.2 tcl, 90% webmail which has been
running perfectly well under stress and load for months...


Urgent help/suggestions are appreciated ..

Allan Regenbaum
Digital People



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Allan Regenbaum DP

Hi Barry .. the machines are decoupled they are in no way dependent upon
each other .. I just mentioned that to show that teh setup works perfectly
on one machine and is failing on another identical config ... seems that as
the production is loade dup something is causing one of our sheculed procs
to not return control of the thread, thereby hanging all the other scheduled
procs. ?

-Original Message-
From: AOLserver Discussion [mailto:[EMAIL PROTECTED]]On Behalf
Of Barry Books
Sent: Tuesday, October 30, 2001 9:14 AM
To: [EMAIL PROTECTED]
Subject: Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS


Did you turn on clustering? I think tasks only run on the master in a
cluster.

Barry Books

-Original Message-
From: Allan Regenbaum DP [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, October 30, 2001 8:03 AM
To: [EMAIL PROTECTED]
Subject: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS


We are running a production AOLServer 3.3.1 + ad13 on solaris
We have 2 machines, duplicates of each other.

Machine A suddenly, without error messages stops running scheduled procs !!
Machine B continues to perform flawlessly
The machine is running acs classic 4.2 tcl, 90% webmail which has been
running perfectly well under stress and load for months...


Urgent help/suggestions are appreciated ..

Allan Regenbaum
Digital People



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Michael Roberts

I have *always* had a problem with scheduled procs, to the point where I now
simply use a cron job and curl to tell AOLserver to do something

Michael

Allan Regenbaum DP wrote:

> We are running a production AOLServer 3.3.1 + ad13 on solaris
> We have 2 machines, duplicates of each other.
>
> Machine A suddenly, without error messages stops running scheduled procs !!
> Machine B continues to perform flawlessly
> The machine is running acs classic 4.2 tcl, 90% webmail which has been
> running perfectly well under stress and load for months...
>
> Urgent help/suggestions are appreciated ..
>
> Allan Regenbaum
> Digital People



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Jim Wilcoxson

Are you exec'ing anything in the scheduled proc?  I've seen some
weirdness in the past with execs getting hung, although not much
recently.  I think it was more of an OS (Linux) issue.

We've used scheduled procs for years w/o any problems on all of our
servers, though we only have a handful (of scheduled procs), like 10.
whereas I've read that ACS has lots of them.

A suggestion would be to run more complex stuff in detached threads
that you launch from the scheduled proc rather than in the scheduled
proc itself.  Then if it screws up, you won't tie up the scheduler
thread.

Jim

>
> I have *always* had a problem with scheduled procs, to the point where I now
> simply use a cron job and curl to tell AOLserver to do something
>
> Michael
>
> Allan Regenbaum DP wrote:
>
> > We are running a production AOLServer 3.3.1 + ad13 on solaris
> > We have 2 machines, duplicates of each other.
> >
> > Machine A suddenly, without error messages stops running scheduled procs !!
> > Machine B continues to perform flawlessly
> > The machine is running acs classic 4.2 tcl, 90% webmail which has been
> > running perfectly well under stress and load for months...
> >
> > Urgent help/suggestions are appreciated ..
> >
> > Allan Regenbaum
> > Digital People
>



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Peter M. Jansson

Make sure that you explicitly release database handles and ns_sets in
scheduled procs.  A connection thread has a bunch of cleanup that it does,
and none of it gets done from a scheduled proc thread.

Pete.



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Jim Wilcoxson

Is this true of detached threads too - no cleanup?  We're probably fine,
but nice to know when to be extra careful.-Jim

>
> Make sure that you explicitly release database handles and ns_sets in
> scheduled procs.  A connection thread has a bunch of cleanup that it does,
> and none of it gets done from a scheduled proc thread.
>
> Pete.
>



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Peter M. Jansson

On Tue, 30 Oct 2001, Jim Wilcoxson wrote:

> Is this true of detached threads too - no cleanup?  We're probably fine,
> but nice to know when to be extra careful.-Jim

I believe so, but it could just be my paranoia.



Re: [AOLSERVER] URGENT AOLSERVER + ACS + SOLARIS

2001-10-30 Thread Rob Mayoff

+-- On Oct 30, Allan Regenbaum DP said:
> Machine A suddenly, without error messages stops running scheduled procs !!

The most likely cause is some scheduled proc that doesn't ever finish.
That proc will tie up the scheduler thread, so the scheduler will never
get a chance to run other scheduled procs.