> I have to add that I quite agree to this. Alternatives to Torque would be quite welcome, and Condor and Slurm look, from eagle's sight, promising.
> Anyways, as I understand, you could potentially replace torque+maui for torque+moab (pricey) and *probably* solve many of the scalability issues. Though I don't know for sure.
From what I've understood the core of moab and maui is quite the same and the differences come in some scale and in lots of additional reporting and tools to manage stuff. So if maui hiccups on some things for no apparent reason you can be pretty sure that with high probability you'll step in the same pile of **** with moab. At least that's what I've understood... We already run a customized torque and maui that have redefined a lot of the limit numbers so that it doesn't entirely die when we get 15k jobs, but at the same time I've had tens of situations where torque+maui goes belly up with just a few hunderd jobs if the conditions are right and sometimes recovery means one has to purge the job base causing huge issues with cream and general accounting etc...
Mario Kadastik, PhD
Researcher
---
"Physics is like sex, sure it may have practical reasons, but that's not why we do it"
-- Richard P. Feynman
|