I was able to create a reproduce of some kind of fluxion performance problem via the following sharness test, which creates 16K fake resources split into 3 different sets via properties, drains 10 of them, then submits various sets of mock execution jobs to them.
In my testing, Fluxion appears to take a long time to initialize from the configured fake reosurces. Then the first set of jobs get 0.4job/s throughput. After that, performance seems to improve. This may be nothing or I may have made some kind of setup mistake, but I thought I'd share just in case.
I dropped this test into t/t8888-scheduling-test.t (results below)
I was able to create a reproduce of some kind of fluxion performance problem via the following sharness test, which creates 16K fake resources split into 3 different sets via properties, drains 10 of them, then submits various sets of mock execution jobs to them.
In my testing, Fluxion appears to take a long time to initialize from the configured fake reosurces. Then the first set of jobs get 0.4job/s throughput. After that, performance seems to improve. This may be nothing or I may have made some kind of setup mistake, but I thought I'd share just in case.
I dropped this test into
t/t8888-scheduling-test.t
(results below)