20010822: LDM error after Linux kernel upgrade

>From: Unidata Mcidas Account <address@hidden>
>Organization: UPRM
>Keywords: 200108221940.f7MJef108307 LDM pqact Linux kernel upgrade

Luis,

>Hi, yesterday I installed the latest Linux kernel on our machine.
>After I restarted it seems that LDM is not receiving the latest images.  I
>checked /var/data/ldm/log/ldmd.log and I got the following error message:
>
>mcidas@atmos:/var/data/ldm/logs$ tail -f ldmd.log
>Aug 22 19:43:10 atmos pqact[203]: child 30168 exited with status 127
>Aug 22 19:43:10 atmos pqact[203]: child 30170 exited with status 127
>Aug 22 19:43:10 atmos pqact[203]: child 30172 exited with status 127
>*snip*
>
>What's the problem?

This is telling us that pqact is exiting prematurely for some reason.
My first guess would be that shared libraries needed by pqact are
either no longer available, or are in a different location.

The first thing to try is:

<login as 'ldm'>
cd bin
ldd pqact

This will show you the list of shared libraries that pqact needs and
finds to run.  If one is needed but not found, the listing will tell
indicate this.

>Is it a damaged queue?

Probably not, but it depends on how you shut down the system before
doing the upgrade.  For instance, if you threw the new kernel in
and then did a quick reboot _without shutting down the LDM_, you may
have damaged the queue.

>I wanted to ask and be sure
>before deleting/remaking the ldm queue.

It was prudent to ask.

>It seems that the pqact child
>processes have no new information to work with.

They are certainly exiting with a bad status.  The fact that they are
being run, however, indicates that there is something in the queue.

>Do you need more information?

If you did a quick reboot, you may have damaged the queue.  In this
case, go ahead and delete the queue and remake it.  This will cause
the entire last hour's worth of data to be sent from your upstream
host(s), but it will be the quickest way of seeing if things improve.

On a different topic, I attended the AMPATH workshop held in Miami,
FL last week.  During the workshop, we were told that Puerto Rico
just got connected to high speed fiber cable.  Figuring that UPRM's
connectivity may improve because of this network upgrade, I tried
doing some IMGDISPs from atmos.uprm.edu back to my home machine.  The
connectivity seemed the same as before the supposed new fast connection.
Can you shed any light on what may be going on networkwise in Puerto
Rico?  If you can't, perhaps Roy Armstrong could (he was at the 
workshop also).

>Thanks,

Please let me know the results of deleting and remaking the queue.

Tom

>From address@hidden Wed Aug 22 17:46:09 2001
>Subject: Re: 20010822: LDM error after Linux kernel upgrade


Tom,
Deleted the queue, remade and it worked fine.  I thought it did an
ldmadmin stop when it when on runlevel 6, I'll add that later today.

re: UPRM networking

   I don't think that upgrade will change much with the university's
connection.  The problem is not the island's connectivity, it's with the
university.  Here's our problem:  The university's main campus is in San
Juan, all the other campuses connect to San Juan(about 12 total).  In our
specific case, UPRM is connected to san juan using a T1(1.45Mbit, right?)
line(yeah, just one).  Actually, there are about 20 of those, but all
but two are used for telephony, the other one is a backup inernet line.
 Ok, this line is rented from a local wireless telecom company(Cingular,
which used to be Cellular One).  It just plain sucks.  All the other
campuses connect to San Juan also, there, the university has the main line
to the inernet.  This line, if I'm not mistaken is a T3(45Mbits), but I'm
not sure, it could be a partial line.  Anyway, there's a HUGE bottleneck
in San Juan, and the university does nothing about it.
   I happen to know a couple of people in charge of the inernet connection
locally.  They're working on an internet 2 line which will connect the
campuses via ATM at OC-12 speed(I could be wrong about that) and they also
plan to add two extra T1's just for our campus.  The internet 2 line is
supposedly finished already(not really sure) and the two T1's is just
something I heard a couple of months ago, may or may not be true.
   Going back to the beginning of the email, the problem is not with the
island's connectivity.  I know a several people who have DSL and have a
fairly good internet connection.  Two weeks ago I got a cable modem, just
to give you an example, downloads are anywhare from 50KB/s to 400KB/s.
The fastet I've gotten from the university's connection is about
150KB/s.  That was at night while I was working on a project, and I
was downloading something from sunsite.unc.edu.  I've got a better inernet
connection home than than the university(at least our campus).
   It used to be worse, it got a lot better when they started filtering
traffic.  (Napster/Gnutella/etc).
   I was just speaking with Amos(my boss at the climate center) today and
we were discussing alternate internet access just for the climate center.
He was leaning towards sattelite access, althought I've heard only bad
things about it.  Another option is reaching an agreement with the cable
company so we can connect throught them.  We'll have to see if we can work
something out.  Hope this clears up any doubts.

Luis Munoz
System Administrator
Caribbean Climate Center @ UPRM