Engine tries submission to resources even when their max capacity has been reached already

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Pick a resource with a low `max_cores` value (e.g., `localhost`)
2. Use a SessionBasedScript (or any `Engine` instance) to submit jobs to that 
resource, using a `max_running` parameter *slightly* larger than the resource's 
`max_cores`.
3. Watch the `Engine` re-try submitting jobs to the resource until its own 
internal list is over.

What is the expected output? What do you see instead?

I would expect the Engine to re-try for `max_running - max_cores` times then 
stop.
Instead, I get a lot of these error messages:

    gc3.gc3libs: ERROR: Ignored error in submitting task 'Application.55408': LRMSSubmitError: Resource localhost already running maximum allowed number of jobs (increase 'max_cores' to raise).

The root cause of this error is likely the following: the `localhost`
resource raises a generic `LRMSSubmitError`, so the `Engine` cannot
infer that the resource is full and should no longer be used for
submission.

There are two things to fix for this issue:

- defined a `CapacityExceeded` exception for resources, raised when no more 
jobs can (ATM) be submitted;

- provide a mechanism for Engine to *temporarily* disable a recource (i.e., 
skip it in the current submission cycle; re-try during the next one).

Original issue reported on code.google.com by riccardo.murri@gmail.com on 5 Nov 2012 at 3:25

[GoogleCodeExporter]

Hi Riccardo,

I'm running into the same problem at the moment, so I assume this issue is 
still unsolved?

Thanks,
Niko

Original comment by imcf.n...@gmail.com on 16 Jul 2015 at 8:06

[GoogleCodeExporter]

Actually, I'm even seeing the errors with a resource configured for max_cores=2 
and max_running=1...

Original comment by imcf.n...@gmail.com on 16 Jul 2015 at 8:35