From aldavis at wi.mit.edu  Thu Nov  4 16:52:56 2004
From: aldavis at wi.mit.edu (Al Davis)
Date: Thu, 04 Nov 2004 16:52:56 -0500
Subject: [CSBi-HPC] NE47 (500 Tech Sq) Power Outage on Saturday
Message-ID: <418AA4B8.5010408@wi.mit.edu>

I just learned today there will be a power outage for the entire 500 Tech Sq 
building this Saturday that will last several hours.  While there is emergency 
power to keep the computers running, there will not be any AC in the server 
room.  Since the power outage will last several hours, I'll be shutting down all 
the BioImaging servers & storage and the CSBi cluster to prevent them from 
overheating.

I'll be shutting everything down on Friday after 5 PM and will start bringing 
all the systems back up starting Sunday morning.  I'll send around another email 
when complete.

Please stop all jobs running on these systems prior to Friday night.  Sorry for 
the late notice as I was not on the original distribution list.

al
-- 
Al Davis                           aldavis at wi.mit.edu | aldavis at mit.edu
Systems Manager                    617.324.0519
CSBi & WI/MIT BioImaging Center    NE47 Rm 311 (500 Technology Sq)

From aldavis at wi.mit.edu  Sun Nov  7 13:49:19 2004
From: aldavis at wi.mit.edu (Al Davis)
Date: Sun, 07 Nov 2004 13:49:19 -0500
Subject: [CSBi-HPC] BioImaging IBM cluster status
Message-ID: <418E6E2F.9090600@wi.mit.edu>

The BioImaging half of the cluster is available for normal use.

al
-- 
Al Davis                           aldavis at wi.mit.edu | aldavis at mit.edu
Systems Manager                    617.324.0519
CSBi & WI/MIT BioImaging Center    NE47 Rm 311 (500 Technology Sq)


From virnau at MIT.EDU  Fri Nov 19 16:46:14 2004
From: virnau at MIT.EDU (Peter Virnau)
Date: Fri, 19 Nov 2004 16:46:14 -0500
Subject: [CSBi-HPC] question
Message-ID: <1100900774.419e69a64436c@webmail.mit.edu>

Hi,

I have a question concerning the IBM clusters. Has anyone figured out how to
write a script, which calls itself (or another script) once it is completed? I
have a job, which needs to run for a couple of days and I don't want to restart
it manually twice a day. Your help would be appreciated!

With best wishes, Peter.




From virnau at MIT.EDU  Fri Nov 19 16:46:15 2004
From: virnau at MIT.EDU (Peter Virnau)
Date: Fri, 19 Nov 2004 16:46:15 -0500
Subject: [CSBi-HPC] question
Message-ID: <1100900775.419e69a72bd02@webmail.mit.edu>

Hi,

I have a question concerning the IBM clusters. Has anyone figured out how to
write a script, which calls itself (or another script) once it is completed? I
have a job, which needs to run for a couple of days and I don't want to restart
it manually twice a day. Your help would be appreciated!

With best wishes, Peter.




From chet at watson.ibm.com  Fri Nov 19 22:51:07 2004
From: chet at watson.ibm.com (Chet Murthy)
Date: Fri, 19 Nov 2004 22:51:07 -0500
Subject: [CSBi-HPC] question 
In-Reply-To: Your message of Fri, 19 Nov 2004 16:46:15 EST.
             <1100900775.419e69a72bd02@webmail.mit.edu> 
Message-ID: <200411200351.iAK3p7lR004264@localhost.chet.org>


Peter,

(\f.\x.f(xx))(\f.\x.f(xx))

*grin*

Actually, you should be able to write a multi-step LoadLeveler job,
whose last step resubmits the same job (one assumes, with different
parameters, but perhaps not ;-)

Look at job steps in the LoadLeveler docs.  It's rudimentary, but
quite nice.  Arbitrary DAGs work.

--chet--