Queue history and some helpful statistics in Gurobi Cluster Manager
AnsweredHi all,
I was wondering whether it is possible to retrieve queue history in Gurobi Cluster manager.
Would be very helpful when checking if computing power is sufficient (or needs to be increased), compute server settings are optimal or tasks waiting to long in queue. The following infos would be helpful:
- state of queue: how much jobs have been waiting in queue and when)
- per job: how long was their waiting time in queue?
- some statistics about the state of all computing nodes: how much jobs have been run, in total, per node, at what time, concurrently?
Thanks.
Regards
-
Official comment
This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum. Or why not try our AI Gurobot?. -
Hi Daniel,
yes, you can see this information either directly or you can infer it from the information available. Specifically, you can see "Waiting time before started" in the web UI of the Cluster Manager in the History tab. However, I think for thorough analysis I suggest you use the REST API: the `/history/jobs` GET request provides (among other things) the items listed below ("startedAt", "queuedAt", "queueIndex").
In terms of the nodes, you can retrieve (again through the REST API) the "NodeMetrics", where one of the items is "processedJobCounter", i.e. you can see how many jobs in total have been processed by that node. But I think if you want a differentiated view by time then I suggest you set up some automated monitoring using the REST API based on your specific KPIs.
Maybe it makes sense to have a call and discuss this topic in more detail? I'll drop you an email and then we can talk if you are interested.
Best regards
Richard
0
Post is closed for comments.
Comments
2 comments