Hi Ian,
On Thu, 9 Jan 2003, Ian Stokes-Rees wrote:
> > > 4) Once the WN is configured, how do I actually go about testing that the
> > > CE, SE, WN, and UI are working? I thought I saw a page explaining how to
> > > do this, but I can't find it now -- do I simply use my UI node to submit a
> > > job and specify my CE as the only target?
> >
> > The most basic tests are just
> >
> > uihost> globus-job-run cehost /bin/pwd
> > uihost> globus-job-run cehost/jobmanager-pbs /bin/pwd
> > uihost> globus-job-run sehost /bin/pwd
>
> Executing each of these results in the following message on the UI:
>
> GRAM Job submission failed because authentication with the remote server failed (error code 7)
>
> while on the host (either CE or SE) the message:
>
> date host GRAM gatekeeper[proc#]: LCAS failed authorization
>
> appears.
On the CE a few things to check.
o Is the gatekeeper running
ps auxwww | grep gatekeep
there should one gatekeeper process.
o Anything in /var/log/globus-gatekeeper.log
o Did you create the gridmapdir lock files as described in
the post install notes.
>
> > uihost> ldapsearch -x -H ldap://cehost:2135 -b 'Mds-vo-name=local,o=Grid'
> > uihost> ldapsearch -x -H ldap://sehost:2135 -b 'Mds-vo-name=local,o=Grid'
>
> These both work just fine, so at least something is installed properly!
Good.
>
> > These are good places to start. Do you have an account on
> > gppui.gridpp.rl.ac.uk , part of the CSF, this is very useful
> > for testing in the first instance.
> > http://atlas.rl.ac.uk/csf/csfuserguide/usrreg.shtml
>
> I do, but it looks like they have an older version of Globus installed.
> Anyway, I'll do some experimenting there.
Did you use gppui.gridpp.rl.ac.uk
This is an EDG CE, v1.4.0 , soon to be v1.4.2 (released 10 minutes ago).
>
> > > Isn't there any way to test it "locally" without having to involve central
> > > (i.e. off-site) replica catalogues and resource brokers? I'd like to be
> > > fairly confident that it is working correctly before getting it tied in to
> > > other points in the test bed.
> >
> > This is a sensible idea of course.
> >
> > You can submit some jdl through a RB with out actually having your
> > machines registered.
> >
> > dg-job-submit -r cehost.ox.ac.uk/jobmanager-pbs-workq helloworld.jdl
>
> I am having another problem with PBS, which is that it doesn't look like
> I've set it up properly. In step 7. of section 7.4 in the PBS
> configuration notes at:
>
> http://www.lnl.infn.it/datagrid/wp4-install/testbed-lcfg/testbed-lcfg.html
>
> I try executing:
/usr/pbs/sbin/qserver -t create
/sbin/service pbs start
/usr/pbs/bin/qmgr </usr/spool/PBS/pbs_server.conf
>
> /usr/pbs/bin/qmgr </usr/spool/PBS/pbs_server.conf
>
> but I get the error:
>
> Unknown Host.
> qmgr: cannot connect to server
>
> No "qmgr" commands produce anything except this output.
>
> I also get an error during updaterpms on my SE node from LCFG:
>
> edg_se_query_1.0.5-1: couldn't find header file
> Can't find original rpm for edg_se_query
Remove edg_se_query from
/opt/local/linux/6.2/rpmcfg/StorageElement-rpm.h
/opt/local/linux/6.2/rpmcfg/UserInterface-rpm.h
This is fixed more generally with edg > 1.4.0.
Steve
>
> Cheers,
>
> Ian.
>
> --
> Ian Stokes-Rees [log in to unmask]
> Particle Physics, Oxford http://www-pnp.physics.ox.ac.uk/~stokes
>
--
Steve Traylen
[log in to unmask]
http://www.gridpp.ac.uk/
|