Huw Lynes wrote:
On Wed, 2009-04-29 at 11:21 -0700, John Bushnell wrote:
Another reason for not having UPS power for an entire cluster is
cooling. In our server room, if we lose regular power, we lose our
large AC units as well. So we need to be careful what we keep plugged
into our emergency power circuits (a relatively tiny amount of
equipment). Having an entire stack of nodes churning away during a
power outage would kill us pretty quickly.
The alternative approach is to UPS everything. Which is what we do.
Although we only have a short run-time installed which will bridge us to
the diesel generator.
In the event that the diesel doesn't kick-in the cluster will cleanly
shut itself down and wait for the UPS to tell it everything is back to
normal (or until an admin intervenes, whichever comes first).
I would put a strong agreement on to this comment. There are
significant benefits to performing an orderly shutdown, then having your
nodes not restart on power restoration. If you don't have a GOOD
mechanism to keep them up through a power problem, it's worth the peace
of mind to babysit them during a restart, regardless of how inconvenient
it is.
gerry
--
Gerry Creager -- gerry.crea...@tamu.edu
Texas Mesonet -- AATLT, Texas A&M University
Cell: 979.229.5301 Office: 979.458.4020 FAX: 979.862.3983
Office: 1700 Research Parkway Ste 160, TAMU, College Station, TX 77843
_______________________________________________
Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
