LOV3, LOV4, LOW1, Foxconn, Ari power outage again
Predrag Punosevac
predragp at cs.cmu.edu
Thu Mar 10 16:19:51 EST 2016
Dear Autonians,
LOV3, LOV4, LOW1, Ari, and Foxconn are out of business again. These are
our five largest CPU computing nodes. I hate to name the names but here
it comes. His name is Ankit Laddha and he did that by running the 30
instances of the same script across the five machines which caused
original problems yesterday.
I thought I was clear in my earlier message that we had only temporary
fix for the electricity problem. Let me now give you the full
explanation. We have old 20 A PDU plugged into 15 A 120V UPS. Dr.
Barnabas has just purchased 20A 208V PDU. However we will need now to
purchase 208V 20A UPS listed here for $1500
http://www.apc.com/shop/us/en/products/APC-Smart-UPS-3000VA-RM-2U-LCD-208V/P-SMT3000RMT2U
Unfortunately I didn't check my notes before the purchase and wishfully
assumed that the UPS donated to us couple of months ago was 208V (which
is the new standard). It is not. It is an old 15A 120V UPS. So until
the new UPS is purchased we will have to be little bit more careful how
we conduct ourselves now when we know that the circuit can be
reproducibly crashed by large CPU loads.
Mr. Laddha has a yellow card now and if he breaks the circuit again I
will have no other option but to temporary suspend his lab account.
Best,
Predrag Punosevac
More information about the Autonlab-users
mailing list