search

kbaseattic / assembly

An extensible framework for genome assembly.
MIT License
12 stars 14 forks source link

Running jobs monitor / User job limit #217

Closed cbun closed 10 years ago

cbun commented 10 years ago

The control server now keeps track of running jobs in realtime, checking for a heartbeat from the compute server. Stale jobs are purged at a given frequency arast.conf.monitor.running_job_freq.

Users are limited to queue+run arast.conf.monitor.running_job_limit. (default 10)

Current counts of the system can be seen via REST:

/admin/system/jobs/
levinas commented 10 years ago

Cool. Please add a comment when this gets deployed on dev.

What's the unit for running_job_freq=43?

If the user job limit is reached, the idea is the client code will print the HTTPError(403, "User Job limit reached") it should be able to catch?

cbun commented 10 years ago

The unit is seconds. I just made it some arbitrary number that's larger than the compute's heartbeat (29). The client should catch, but this isn't implemented yet.

levinas commented 10 years ago

Ok. I think the client already catches cherrypy HTTP exceptions with the message.

On Aug 12, 2014, at 10:06 PM, Christopher Bun notifications@github.com wrote:

The unit is seconds. I just made it some arbitrary number that's larger than the compute's heartbeat (29). The client should catch, but this isn't implemented yet.

— Reply to this email directly or view it on GitHub.

cbun commented 10 years ago

Please add a comment when this gets deployed on dev.

Deployed ar-ctrl-edge/bigmem