> Am 30.05.2017 um 23:49 schrieb Roberto Nunnari <[email protected]>: > > Dear Reuti. > > Thank you very much for your help. I very much appreciate it. :-) > > After some trouble (getting the right architecture and the man pages) I > managed to build it all (I hope). > > I guess "sh scripts/bootstrap.sh && ./aimk" will build all with berkeleydb > and not -spool-classic, right?
Yes. But unless you have a really large cluster, there is no big difference. Some even prefer classic-spooling, as you can read all text files without any extra tools. > I'll let you know tomorrow how it goes. > > But I'm new to arc and that will be a bit of a challange. :-) > Can I add it later or I need to do it together with SoGE? You mean ARCo for the accounting? As I wrote in the other eMail: we never used it, as we never had a need for it. There was the idea to use it, but essentially the pending queue is so short, and there are only short phases where some waiting time exists, that we skipped this. The only "accounting" we use, is the accounting field in the submission command `qsub -A any_string …` (more or less a comment) where my job generator will put in the name of the used application. Neither the PI nor I are interested who used which application how often. But the overall absolute number of each quantum chemistry application we have to know per timeframe to order the correct number of licenses (and some applications uses license daemons which controls this). Later on something like: $ qstat -j "*" | grep account | sort | uniq -c provides the necessary information. That's all we use. -- Reuti > > Thank you and best regards. > Roberto > > > > Il 30.05.2017 19:07, Reuti ha scritto: >> >>> Am 30.05.2017 um 18:12 schrieb Roberto Nunnari <[email protected]>: >>> >>> >>> >>> On 30.05.2017 17:19, Roberto Nunnari wrote: >>>> Anyways.. I built Soge 8.1.9 with >>>> cd source >>>> sh scripts/bootstrap.sh && ./aimk >>>> and now? how will I install that? it scattered the files in several places >>>> under source.. Is there a make install? >>> >>> Sorry to insist.. but really.. I need an answer to that question, as it >>> will be the first thing I'll need tomorrow morning. >>> >>> Thank you for understanding and best regards. >> >> There is a file README.BUILD There it's outlined how to proceed depending on >> whether you want a direct install, or transfer it to another machine (most >> likely). >> >> For the second: create the staging directory like $HOME/MY_SGE, set the >> environment variable SGE_ROOT to it, advice the "scripts/distinst -local >> -allall lx-amd64" to copy all relevant files thereto. (In case it's the same >> machine where you proceed: reset SGE_ROOT to /usr/sge or alike for the final >> target.) >> >> This staging directory is your personal distribution, which you can tar >> (while being inside this directory) and transfer it to any machine/cluster >> and use it for installation: >> >> tar czf ~/my-sge-8.1.9.tgz * >> >> The staging directory itself you don't need any longer. Untar your >> distribution tarball inside the final target /usr/sge >> >> Inside the tarred/untarred directory run as root: ./install_qmaster >> >> For the exechosts it's not really necessary to install anything. The >> install_execd will just add a particular host to the SGE installation and >> install the /etc/init.d/sgeexecd on it. When cloning the exechost, it's just >> necessary to add them to the known machines in SGE and start the sgeexecd >> thereon. >> >> -- Reuti >> _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
