Trying to get the MDS (GIIS-GRIS) server up and the globus-condor interface working: Everything is done as user globus, unless otherwise noted. globus install path is /usr/local/globus, globus deploy path is /opt/globus. In /usr/local/globus/etc/grid-info-hosts.conf, add: ouhep2.nhn.ou.edu /usr/local/globus/services/i686-pc-linux-gnu/bin \ ouhep1.nhn.ou.edu - This should configure the GIIS/GRIS servers during setup and deployment, but it's not working correctly yet, so maybe it's wrong. See at the end. In /usr/local/globus/etc/globus-services.conf, add: ouhep1.nhn.ou.edu jobmanager-condor stderr_log,local_cred - \ ${libexecdir}/globus-jobmanager globus-jobmanager \ -conf ${sysconfdir}/globus-jobmanager-condor.conf -type condor \ -rdn jobmanager-condor -machine-type unknown This will configure the condor interface during deployment. Then run globus-setup: [globus@ouhep1 globus]$ /usr/local/globus/sbin/globus-setup M D S : S E T U P P R O C E D U R E This script will ask some questions about site specific information. The answers to these questions will help to setup your local MDS infrastructure. Pressing RETURN in response to such a question will enable the default. This script will overwrite the file: /usr/local/globus/etc/grid-info.conf Do you wish to continue (y/n) [y] : ======================================================================== (1) Hostname for site's organization server: [ ouhep1.nhn.ou.edu ] (2) Port number for site's organization server: [ ] (3) DN for the organization: [ dc=nhn, dc=ou, dc=edu, o=Grid ] (4) Save settings (5) Quit ======================================================================== (h)elp ======================================================================== 2 Enter the "port number" for your site's MDS server [ ] : 2136 ======================================================================== (1) Hostname for site's organization server: [ ouhep1.nhn.ou.edu ] (2) Port number for site's organization server: [ 2136 ] (3) DN for the organization: [ dc=nhn, dc=ou, dc=edu, o=Grid ] (4) Save settings (5) Quit ======================================================================== (h)elp ======================================================================== 4 ======================================================================== (1) Hostname for site's organization server: [ ouhep1.nhn.ou.edu ] (2) Port number for site's organization server: [ 2136 ] (3) DN for the organization: [ dc=nhn, dc=ou, dc=edu, o=Grid ] (4) Save settings (5) Quit ======================================================================== (h)elp ======================================================================== 5 G S I : C O N F I G U R A T I O N P R O C E D U R E Before you use the Grid Security Infrastructure, you should first define the DN (distinguished name) that should be used for your organization's X509 certificates. If you do not define a DN, a default DN will be assigned to you. This script will ask some questions about site specific information. This information is used to configure the Grid Security Infrastructure for your site. For some questions, a default response is given in []. Pressing RETURN in response to such a question will enable the default. This script will overwrite the file -- /usr/local/globus/etc/grid-security.conf Do you wish to continue (y/n) [y] : ======================================================================== (1) Base DN for user certificates [ ou=nhn.ou.edu, o=Globus, o=Grid ] (2) Base DN for host certificates [ o=Globus, o=Grid ] ======================================================================== (q) save, configure the GSI and Quit (c) Cancel (exit without saving or configuring) (h) Help ======================================================================== q The setup is complete. [globus@ouhep1 globus]$ Now stop globus services, backup old /opt/globus, make /opt/globus_save/etc/globus-gatekeeper.key world readable, and create a new /opt/globus and give ownership to globus: # /etc/rc.d/init/globus stop # mv /etc/globus /etc/globus_save # chmod a+r /opt/globus_save/etc/globus-gatekeeper.key # mkdir /etc/globus # chown globus.globus /etc/globus Then run globus-local-deploy: /usr/local/globus/sbin/globus-local-deploy \ -cert /opt/globus_save/etc/globus-gatekeeper.cert \ -key /opt/globus_save/etc/globus-gatekeeper.key /opt/globus Afterwards, follow instructions in /opt/globus/tmp/globus-root-instructions. Don't forget to read-protect the new /opt/globus/etc/globus-gatekeeper.key again, otherwise globus will not start up: # chmod go-r /opt/globus/etc/globus-gatekeeper.key Now, modify several condor files: In /opt/globus/etc/globus-jobmanager-condor.conf: Remove # Edit the following two lines to complete # the configuration of your condor jobmanager (not strictly necessary, but you'll get lots of warnings in the log file if you don't, since whichever script uses this file doesn't recognize #'s and complains that it will ignore those lines, one word at a time!) and change -condor-arch -condor-os to -condor-arch INTEL -condor-os LINUX Warning: case sensitive! And now, and this is not written anywhere - at least I didn't find it - in /opt/globus/etc/i686-pc-linux-gnu/globus-sh-commands.sh, set the following condor commands: GLOBUS_SH_CONDOR_Q="/usr/local/bin/condor_q" GLOBUS_SH_CONDOR_RM="/usr/local/bin/condor_rm" GLOBUS_SH_CONDOR_STATUS="/usr/local/bin/condor_status" GLOBUS_SH_CONDOR_SUBMIT="/usr/local/bin/condor_submit" Now start up globus again: # /etc/rc.d/init/globus start Now you can submit jobs to the condor queue with # globus-job-run ouhep1.nhn.ou.edu/jobmanager-condor Or # globus-job-submit ouhep1.nhn.ou.edu/jobmanager-condor which will print out a for you; something like https://ouhep1.nhn.ou.edu:1475/1334/986942726/ The problem with globus-job-submit is that when you're trying to get the output with #globus-job-get-output it complains about not finding globus-gass-cache: /usr/local/globus/tools/i686-pc-linux-gnu/bin/globus-sh-exec: \ /usr/local/globus/tools/i686-pc-linux-gnu/bin/globus-gass-cache: \ No such file or directory I can't find globus-gass-cache anywhere in the globus areas, so I don't know what's going on here. Any ideas? Also, as I already pointed out earlier, the GIIS/GRIS servers still don't work right, the /opt/globus/var/globus-system.log file complains about: Mon Apr 9 18:21:08 GMT 2001 globus-daemon-monitor[17949] Error: \ globus-hbm-client-register not found. Mon Apr 9 18:21:08 GMT 2001 globus-daemon-monitor[17949] slapd (pid 18126) died And I have no idea what that means. Ideas?