Bug #12834
Updated by Ward Vandewege over 6 years ago
In a containers api only scenario, the crunch_worker_state field on the node record is never updated.
<pre>
# sinfo
PARTITION AVAIL TIMELIMIT NODES STATE NODELIST
compute* up infinite 63 drain* compute[4,6-67]
compute* up infinite 445 down* compute[0,68-511]
compute* up infinite 4 alloc compute[1-3,5]
crypto up infinite 63 drain* compute[4,6-67]
crypto up infinite 445 down* compute[0,68-511]
crypto up infinite 4 alloc compute[1-3,5]
# RAILS_ENV=production bundle exec rails c
2.3.0 :003 > Node.where(:slot_number => 1).first.crunch_worker_state
=> "down"
</pre>
This means we can't show busy/idle nodes in Workbench, the values are always zero.