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

19990121: Question about bootup

>From: address@hidden
>Organization: University of Northern Iowa
>Keywords: 199901212112.OAA21273  ldm queue reboot 

Alan-

>       Am I correct in understanding from your last
>message that Findeisen is presently configured to get
>the last hour's worth of data after rebooting?  You
>mentioned that this bogged down the system last week,
>but apparently it did not when Drew brought Findeisen
>back to my office and started it up.  The reason I'm
>asking is that I'd like to move Findeisen down to our
>student computer lab so that my forecasting class can
>start making use of GARP and NWX, but I'd rather not
>cause problems by shutting it down and them rebooting
>if it's likely to bog
>the campus down again.  If I moved Findeisen earlier
>in the day (say, 8:30 am), is it less likely to cause
>problems since the traffic tends to be lighter on the
>network then?

Since I don't have the root password anymore on your system,
I can't comment out the section in the /etc/rc.d/init.d/ldmd
file that deletes and remakes the queue.  If this is commented
out, then you will not get the slug of data from the past hour.
The following lines could be commented out by putting a # at
the start of the line:

                /bin/su - ldm -c "$LDMBIN/ldmadmin delqueue"
                /bin/su - ldm -c "$LDMBIN/ldmadmin mkqueue"

The only reason this is in there is in case the queue gets corrupted
by a sudden halt of the system (and even that is not a guarantee that
the queue will become corrupted).  So of these were commented out and
you shut down the ldm before halting/rebooting the system (ldmadmin stop)
then you would not get the slug of data.

If there was a power outage and the system rebooted suddenly and
the queue was corrupted, then the effect would be that the ldm would
not start and the reboot of the system might hang.

Early morning is probably not a good time to restart the system if
you don't comment out the lines, because that is when all the model
data are flowing in.  Later afternoon would be better (4:30-5pm).

>       A second issue.  Mid last month, my newest and
>most capable OS/2 box (Solberg) lost its hard drive.
>Melanie's workers verified that for me today;  I had
>hoped it might be something minor; silly me.  Anyhow,
>it appears that it should still be covered by the
>warranty.  I was thinking I probably should have Linux put
>on the new drive and run GARP/NWX/McIDAS-X.  Solberg
>is a Pentium 150 with 48 MB RAM and (assuming the new
>drive matches the old) a 2.5GB hard drive.  Would this
>be adequate?

The only thing that would concern me is that it only has 48 Mb
of memory.  64 (or more) would be better, but it should run in
48.  Since OS/2 is going to be sunset anyway, then you should
move in the Unix direction and this would be a way to do it.
I'm running McIDAS-X at home on a 133 Mhz system with 64 Mb memory
(using ADDE to get data) and it works fine.  We've run Garp/NWX
on this system when it was in the office and it ran fine.

If you are going to do this, then you need to set up the NFS server
on findeisen to serve up data to solberg and have solberg mount
the data directory (/var/data/ldm) on findeisen as /var/data/ldm
on solberg.  If you do that, setting up GARP/NWX should just be a
matter of creating an NAWIPS-5.4 directory in /home/gempak on
solberg and then downloading and unpacking the linux binary from
our FTP server.

You could also mount /home on findeisen as /home on solberg and
then run everything off the NFS mounted drive.  That way you
would not have to install any of the software on solberg - just
create accounts for gempak, mcidas and wxuser - the same as on
findeisen.  You would not need ldm or mcadde acccounts on solberg.
I can talk to Drew or you about the merits/problems with each
setup if you want.  Let me know.

Don