On Thu, 12 Jul 2012, Dave Love wrote:

Mark Dixon <m.c.di...@leeds.ac.uk> writes:

Things I think we've used starter_methods for in the past:

Gosh.  You live in interesting times^Wclusters.

I've certainly had some interesting problems to tackle. Something's got to keep me busy in funding droughts....


...
* Transparent (from the job script's perspective) serial BLCR
  integration

Could you post the recipe/code?  DMTCP is facing the knife for exactly
that, but C++ encourages displacement activities.

Sure, I'll dig it out (to follow under new thread)


* Allocating hardware (e.g. core binding in pre-6.2 versions, graphics hw)

How well does that work without scheduler support?
<https://arc.liv.ac.uk/trac/SGE/ticket/1426>

It's a horrible kludge.

It involved having a simple daemon on each compute node that kept track of what resources was allocated. The starter_method would request a resource and give it a PID to keep track of. Each time it was asked to allocate something, it would first check that all the PIDs were still alive.


* Firewall twiddling

Is that to control access to horrible licence servers, for instance, and
is it possible to do it per-process with iptables?  (We've been somewhat
stymied by old stuff in RH5 and didn't pursue it.)

It was about letting desktops to talk to paraview server processes on compute nodes for graphics rendering.


...
some have elaborate shell
scripts setting environment variables.

More elaborate than
<https://code.google.com/p/slurm-spank-plugins/wiki/UseEnv>, for
instance?

Well, I often set environment variables as a by-product of what my scripts are doing. So it's more about whether the language is flexible enough to do all the other things I need it to do to get the information to stick into an environment variable.

The greatest strength of the starter_method is that it lets us do things that gridengine developers haven't anticipated :)

Mark
--
-----------------------------------------------------------------
Mark Dixon                       Email    : m.c.di...@leeds.ac.uk
HPC/Grid Systems Support         Tel (int): 35429
Information Systems Services     Tel (ext): +44(0)113 343 5429
University of Leeds, LS2 9JT, UK
-----------------------------------------------------------------
_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users

Reply via email to