[gridengine users] Having trouble installing SGE on a new execution host
reuti at staff.uni-marburg.de
Mon Jan 2 09:52:13 UTC 2017
> Most of the cases there is no need to "install" anything on an additional exechost when you have already a working cluster.
> [Mun] Really? I was basically trying to follow the old "Sun N1 Grid Engine 6.1 Installation Guide" instructions to install an Exection Host from the following URL:
> - Prepare a proper /etc/hosts or NIS or alike on the new machine, so that all machines in the cluster are known for it (and also the old machines should be able to reference the new one)
> - Mount /opt/sge or /usr/sge on the new exechost
> [Mun] When SGE was initially installed, a common mount was not used. SGE_ROOT is local to each host. It doesn't "feel" right to copy $SGE_ROOT/default from a working host to the new host; but I don't know how to get that directory on the new host otherwise.
Just copy the complete $SGE_ROOT then to the same location as on the other exyechosts to the new node. Especially the $SGE_ROOT/default/common contains the settings how to address the qmaster, i.e. its name. This is the idea to share at least this directory branch from the complete installation.
The spool directory inside $SGE_ROOT/default can be skipped, as it will be recreated during startup of the exechost (i.e. one for this particular node).
If this is a new installation of Linux, it might be necessary to adjust $SGE_ROOT/util/arch, so that also newer Linux kernels are covered.
> - Copy $SGE_ROOT/default/common/sgeexecd to /etc/init.d
> Depending on the startup of services you need either:
> # /etc/init.d/sgeexecd start
> # chkconfig --add sgeexecd
> # systemctl daemon-reload
> # systemctl start sgeexecd.service
> # systemctl enable sgeexecd.service
> BTW: Is tmpdir in the queue definition just /tmp or do you need an additional /scratch or alike on the new machine too?
> [Mun] I don't understand this question, sorry. Are you referring to the SGE queues?
> -- Reuti
More information about the users