RLM License Usage | ||||||
posted by Jesse Kretschmer on March 30, 2017, 7 a.m. | ||||||
| ||||||
Thread Tags: discuss-at-studiosysadmins | ||||||
|
Howdy all, Why does RLM report more licenses in use than are available? I recently started tracking license utilization by parsing the output from license servers command line utils (rlmutil, lmutil, tlm_server). Now that I have a couple weeks of data, I am seeing occasional over-usage. I've monitored the output and I actually have more checkouts listed than available licenses. It makes my graphs look weird. More about my setup: I just setup prometheus with grafana. (It defaults to 15 days of storage, but there are ways to make it keep data for much longer._ I have a python service to read from all the license servers every minute, and I build the metrics like this: licenses_issued{feature="rvsolo *.*.*",server="licenseserver",type="TweakLM",vendor="Tweak"} 99 licenses_inuse{computer="workstation001",feature="rvsolo *.*.*",server="licenseserver001",type="TweakLM" user="jkrets",vendor="Tweak"} 1 licenses_inuse{...,user="someoneelse",...} 1 ... I am querying the data back grouped by computer knowing that only one license is needed per machine even if there are multiple checkouts on that machine. sum(min(licenses_inuse{feature=~"nuke_r.*"}) by (computer)) Perhaps is just the way that the vendor daemon works :/ Any thoughts or suggestions would be welcome. Cheers, Jesse |