[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20030708: LDM-6 upgrade at McGill (cont.)

>From:  Alan Schwartz <address@hidden>
>Organization:  AOS
>Keywords:  200307071420.h67EKGLd015022 LDM-6 setup request

Alan,

>maelstrom2 is back to it notmal allocation of diskspace.

Do you know what was going on with the disappearing disk space
under '/'?

>I tried to restart LDM but I'm not sure if it is working correctly

When I logged off this morning, I had started the LDM and it was
running with no problems.  Did you try to restart it without first
stopping it?  I would check on this, but I can get to maelstrom2 at the
moment.  Did you turn access to maelstrom2 off?

>As to the other points you brought up in your previous email,
>I will have to check with Prof. John Gyakum as to his plans
>for LDM.

OK.  One of the more important things that should be attended to is the
clock drift.  This is easily seen from latency plots for any of the
streams that maelstrom2 is receiving:

http://www.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?HDS+maelstrom2.meteo.mcgill.ca

Another problem I was seeing, but was not sure of a cause was the
inability of maelstrom2 to mail back the .stats files that were being
generated by the 'exec pqbinstats' action in ldmd.conf.  Because of the
disappearing disk space, and since I was still at home at the time, I
thought it wise to simply shut off the generation of the stats and not
try to mail them back to unidata (address@hidden).  For
reference, maelstrom _is_ mailing back its LDM stats with apparently no
problems.

>He is in Japan at the moment and I expect him back next week.

If he is at the same conference as another community member that I was
working with this past weekend, he probably won't be back until the 14th.

Tom

>From: Alan Schwartz <address@hidden>
>Date: Tue, 08 Jul 2003 15:03:37 -0400

>Tom,

>The disk space problem seems to be related to the objectserver program
>It allows some client programs to access system resources.
>As this is a fairly useless piece of software, I have removed it from
>the system.
>(The problem began after the time was changed, but I don't know the
>reason)

>The removal of the objectserver required a reboot.
>This flushed the route tables, which blocked remote logins
>I have reset the tables.
>This, of course, blocked LDM as well.
>So now LDM is working once again and you shoud be able to login.

>As a rule we limit those machines which have the ability to send mail.
>maelstrom2 was never configured to send mail. This has now been done.

>--
>Alan Schwartz
>Department of Atmospheric and Oceanic Sciences
>McGill University
>805 Sherbrooke ST. W.
>514-398-3761
>address@hidden