In response to SIGUSR2, the bebopd will begin logging extra information to the log file (/var/log/cplant relating to it's attempts to find free compute nodes. For non-PBS jobs, the bebopd just indicates whether all the queried compute nodes responded. For PBS jobs, the bebopd lists certain counts. This may be helpful when debugging a situation where compute nodes are not responding to status requests.
Sending another SIGUSR2 to the bebopd will turn off the extra information.
allocate_nodes_for_pbs (8), 1 requests, 128 nodes
got all expected updates
1018 queries
minPct 24 maxPct 1047
226 free 0 pending 0 already in use by PBS job
98 unneeded pcts released
In this example, the bebopd reported that it received a yod request for 128 nodes for a PBS job. The bebopd send 1018 status queries to PCTs and got all expected updates back. It queried active compute nodes in the physical node number range of 24 through 1047. (Not all of those nodes were active compute nodes.) Of those, 226 were available to run a job. The bebopd sent a list of 128 of them back to yod and sent a message to 98 others that they would not be needed. The bebopd also lists a count of pending nodes and nodes already in use by this PBS job. Pending nodes are compute nodes that have been allocated to a yod process but which haven't yet been contacted by yod to run a job.