[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] flocking problem



On Thu, 2005-01-06 at 15:46 -0600, David A. Kotz wrote:
> In my department, I have 3 pools of Condor machines -- 2 dedicated
> computing clusters and the desktop machines.  All 3 pools are configured
> for 2-way flocking  with one another.
> 
> The machines in the Mastodon cluster advertise the attribute
> "InMastodon", and the machines in the Scout cluster advertise the
> attribute "InScout".  They have lines such as "InMastodon = TRUE" in the
> local config files.
> 
> Prior to splitting the machines into 3 separate pools (for the purpose
> of keeping separate logs for reporting to funding agencies), users could
> specify in their submit descriptions that they wanted jobs to run on one
> of the cluster nodes by including something like this, "Requirements =
> InScout".
> 
> Now if a user submits a job from one pool, requesting an execute machine
> in a different pool, the job sticks in the queue because Condor can't
> find a match.
> 
> Is this a configuration issue, or a fundamental limitation?  I'm
> currently running Condor 6.6.6 on Linux.
> 

Mere seconds after I submitted my question, the jobs inexplicably
decided to flock.  Prior to sending, I had watched at least three
consecutive negotiator cycles say that there were no matches for my test
jobs.  Prior to that, I left a job cluster overnight trying to flock in
the opposite direction, to no avail.

I'm guessing that I have a configuration issue or that Condor hates me,
or both.


-- 
David A. Kotz <dkotz@xxxxxxxxxxxxx>