Hi, Am 30.10.2012 um 14:20 schrieb Petter Gustad:
>> Am 30.10.2012 um 10:53 schrieb Petter Gustad: >> >>> Does anybody have a pointer to GRE installation docs for CentOS 6.3? >>> >>> I've been running GRE version 6.2u5p2 (built from source) on Gentoo >>> systems for some time. Now I'm trying to add some new nodes running >>> CentOS 6.3 using the gridengine-6.2u5-10.el6.4.x86_64 package. >> > > >> Just use the version you have already in the shared /usr/sge or your >> particular mountpoint. > > I should probably try this first, at least to verify that it's > working. But later I would like to migrate to the CentOS on all my > exechosts and leave the installation to somebody else. Then it's bets to start with the qmaster. >> For exechosts there is no real installation necessary. It's >> sufficient to add the new exechosts as adminhosts, and then start >> the sgeexecd on the nodes (you might want to install it in >> /etc/init.d/sgeexecd or alike with appropriate links so that they >> start while booting). The script you will find in >> /usr/sge/default/common/sgeexecd. > > I'll try this manual approach. > >> During startup of the sgeexecd they will become exechosts >> automatically in SGE's list of exyechosts. > >> NB: It's not advisable to mix different versions of SGE in a cluster >> (while it's fine to mix different platforms of the the same >> version). > > OK. I will try to get the old version running first, then migrate to > the more recent version as I replace Gentoo with CentOS on the > exechosts. Only the exechost? Then I suggest to reinstall the qmaster on the head node with the version you want to use on all exechosts. >> PS: You installed SGE in addition locally on each node with the >> gridengine-6.2u5-10.el6.4.x86_64 package? It might get confused >> having more than one version of the tools. > > There is not PATH pointing to the old version. When I first tried the > old version was not even mounted. It's quite usual to mount /usr/sge (or alike) and /home in the cluster and have only SGE's spool directory local (e.g. /var/spool/sge): http://arc.liv.ac.uk/SGE/howto/nfsreduce.html -- Reuti > However, I would assume CentOS users with no previous installation > would experince the same problem. > > Thank you again for your helpful reply. > > Best regards > //Petter > > >>> But the installation procedure seem to be somewhat different to what >>> I'm used to. So where can I find an installation guide for the CentOS >>> version? >>> >>> My problem seem to be related to that sge_coshepherd and sge_shepherd >>> is missing. Is this a problem with the CentOS 6.3 package or is there >>> a different installation procedure on CentOS? >>> >>> Here's the error message I get when I try to run inst_sge: >>> >>> >>> # export SGE_ROOT=/usr/share/gridengine >>> # sh ./inst_sge -x >>> missing program >sge_coshepherd< in directory >./bin/lx26-amd64< >>> missing program >sge_shepherd< in directory >./bin/lx26-amd64< >>> >>> Missing Grid Engine binaries! >>> >>> A complete installation needs the following binaries in >./bin/lx26-amd64<: >>> >>> qacct qlogin qrsh sge_shepherd >>> qalter qmake qselect sge_coshepherd >>> qconf qmod qsh sge_execd >>> qdel qmon qstat sge_qmaster >>> qhold qresub qsub qhost >>> qrls qtcsh sge_shadowd qping >>> qquota >>> >>> and the binaries in >./utilbin/lx26-amd64< should be: >>> >>> adminrun gethostbyaddr loadcheck rlogin uidgid >>> authuser checkprog gethostbyname now rsh >>> infotext checkuser gethostname openssl rshd >>> filestat getservbyname qrsh_starter testsuidroot >>> >>> Installation failed. Exit. >>> >>> >>> Thanks! >>> Best regards >>> //Petter >>> _______________________________________________ >>> users mailing list >>> [email protected] >>> https://gridengine.org/mailman/listinfo/users >> _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
