Biowulf Cluster Downtime on June 16/17 (Biowulf)

This email is a follow-up to one sent last week to all
Helix Systems users regarding a scheduled downtime of
all systems on June 16/17.  The Biowulf staff will be
taking advantage of this downtime to perform a number
of hardware and software upgrades.

A provisional schedule for the Cluster downtime follows:

Monday June 16
3 pm: the batch system will be shut down.  All jobs running
on the cluster at this time will be terminated.  Owners of
running jobs will be notified of which jobs were terminated.

Queued jobs will be unaffected and will be scheduled to run
when the system resumes.  Please do not submit large 'swarms'
which are not likely to complete before June 16 3 pm!

6 pm: the login node (biowulf.nih.gov) will be unavailable.

Tuesday June 17
6 am - 9 am: Cluster nodes will be brought online. Queued jobs
will be started when the batch system is brought up. We hope
to have the cluster fully operational and available to users by
9 am.  The login node will also be available at this time.

Please send questions and comments to staff@biowulf.nih.gov.

The original announcement follows:

> From: Helix Systems Staff 
> Date: May 28, 2008 12:23:30 PM EDT
> To: HELIX-USERS@LIST.NIH.GOV
> Subject: Helix Systems Downtime on June 16
> Reply-To: Helix Systems Staff 
>
> All Helix Systems (Helix, Doublehelix and the Biowulf cluster)
> will be unavailable during the evening/early morning of
> June 16-17, during which two major upgrades will be performed:
>
> -- the transition of Helix to a new hardware and software
> platform (from an 8-processor SGI running the IRIX operating
> system to a 16-processor Sun Opteron system running Linux).
>
> -- power feed upgrades in the Helix Systems machine room,
> which will increase power reliability and robustness.
>
> Helix and Doublehelix will become unavailable to users at 9 pm
> on Monday June 16.  We anticipate that the systems will become
> available to users no later than 9 am on Tuesday June 17.
>
> During the downtime, all Helix services will be inaccessible
> including Helix email via POP & IMAP clients, Squirrelmail and
> pine. Email to Helix will be queued during the downtime for
> delivery once the system is back up.
>
> Details regarding the Helix transition will be sent to all users
> in the near future.  Details about the Biowulf cluster downtime
> will be sent to Biowulf users.
>
> We regret any inconvenience this may cause.
>
> Helix Systems Staff.