Hello everyone
I wish to make a few comments about the the upgrade I finally almost
completed, in a regular way.
I've been stuck for a while with unsuccessful job submission : no incoming job
just could make it one it had reached the CE. The logging info never showed
that the job was transferred to the batch system (even when the latter is
present). That puzzled me for a while, and then I recalled that in the
site-config.h, the SITE_GLOBUS_TCP_RANGE has moved from 30000-31000
to 20000-25000.
And that was the cause of the failure, as when I made a try
using the former port range, my jobs were finally accepted and not stuck in
the "Ready" status any more, but had a chance to get "Scheduled" and
eventually to complete.
Indeed, our site is a bit firewalled and port range 20000-25000 are closed
while 30000-31000 are opened.
I do *not* run the firewall, neither have I a simple way to change it's
configuration myself, I have to have it done by someone else.
Hence the question : do I have to get the new port range opened on the
firewall, or can I continue with port range 30000-31000 ?
As our site is quite very small, I suppose we don't need such a wide port
range for job submission ?
Also note that on my WNs I had to have an additionnal flag set in
the ssh_config file saying " EnableSSHKeysign yes " to enable file copy from
the execution node back to the pbs server. LCG realease notes do not mention
it.
Apart from this point, I still have a RGMA issue, but for this I know it's not
a firewall issue as ports 8080 and 8088 are opened in both directions, but I
still hope I will finally get it to work properly. MySQL config seems OK, all
tables are created (though empty) and permissions are granted the way they
should. RGMA was even reported to be OK by the GIIS monitor for our site with
the previous release, altough it wasn't collecting anything. Shall I drop
each and every table and rebuild the DBs from scratch using the provided sql
scripts or is there a tomcat change I'm couldn't point up to now ?Any advice
would be welcome.
Thank you
David Weissenbierbach.
On Sunday 30 January 2005 21:18, [log in to unmask] wrote:
> On Sun, 30 Jan 2005, David WEISSENBACH wrote:
> > > > Concerning pbs packages that disappeared from the CE and the WNs,
> > > > I also finally got them to be installed by lcfg : there was NO
> > > > include of lrms-server-rpm.h and lrms-client-rpm.h in the packages
> > > > list. I fixed it in the LCG-2_2_0 way.
> > >
> > > I do not see how you could find that lrms-server-rpm.h and
> > > lrms-client-rpm.h were not included: CE-rpm and WN-rpm clearly include
> > > them!
> >
> > I finally see : I used the directory rpmlist_rh73 rather than rpmlist
> > The differences between them are exactly what we've been aguing about,
> > with the tank thing as a final addition.
> >
> > [...]
> >
> > I suppose I'd better stick to the rpmlist version ?
>
> Indeed. The "rpmlist_rh73" directory is for the manual/YAIM installation,
> which apparently takes care of pbs and the CAs in a different way.
>
> > > [...]
> > >
> > > Just about every rpm list includes "security-rpm.h", so there is
> > > something funny going on...
> >
> > Well, if it's fun for you, I'll be glad with it :-) The purpose of this
> > wasn't to bore anyone indeed...
>
> (funny == strange)
|