Hi Ake,
Åke Sandgren wrote:
> Having a valid value (4444) as a marker for invalid data is stupid.
> Our cluster have a varying degree of jobs in the system, sometimes as
> many as 12-16000!!
You are right that the current value is a little silly. It's
historical, I didn't see that many waiting jobs very often when the
framework was being built, but it happens regularly at many places these
days. A bug was submitted last week to "upgrade" the value:
http://savannah.cern.ch/bugs/?23444
I apologize for my lack of foresight ... at the time 4444 seemed like a
perfectly reasonable value. At least it was an improvement on the
previous value "0" which was responsible for lots of black holes, and
still seems to be a popular replacement value for those sites that don't
like 4444.
J "waiting for the 7777777 complaints to show up" T
|