6.7. Reinstall All Compute Nodes with SGE

This section describes how to reinstall all the nodes under the control of SGE.

As root, execute:

# /opt/gridengine/examples/jobs/sge-reinstall.sh

This will produce output like:

Your job 1 ("reboot.qsub") has been submitted
Set compute-0-0.local for Reinstallation
Your job 2 ("reboot.qsub") has been submitted
Set compute-0-1.local for Reinstallation
Your job 3 ("reboot.qsub") has been submitted
Set compute-0-2.local for Reinstallation

For each compute node, the script submits a high-priority reinstall job. The reinstall jobs will move ahead of the currently queued jobs (running jobs will not be disturbed). When a busy compute node becomes free, the next job submitted to it will be the reinstall job. All remaining jobs will remain queued until the node completes its installation, after which SGE will recognize the node as up and then submit a previously queued job to it.