How to put IC HEP into downtime
-
Fill out the form in the GOC DB asap.
-
Put the nodes in downtime in nagios.
best to do this by host group: On the left hand side click "Hostgroup Overview",
then e.g. BDII Servers (bdii-group) click on the group in brackets, then
"Schedule downtime for all hosts in this hostgroup".
- Two days before, on the CE: Set /etc/sge-jobmanager/cluster.state to "Draining"
- At start of downtime: Set /etc/sge-jobmanager/cluster.state to
"Closed"
- Two days before: Disable the queues on the CREAMCEs: glite-ce-disable-submission
- Stop the queues in SGE: On sgemaster03 (I can be root now)
do 'qmon', then 'Queue control', click on grid.q and then on 'disable'.
- At the end of downtime: Stop the gatekeeper and use opportunity to clean
out some home directories. Restart the gatekeeper and set
/etc/sge-jobmanager/cluster.state to Production. Enable the queues in SGE (if applicable).