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

19990218: mclap mount problem



>From: Anthony Rockwood - MSCD Meteorology <address@hidden>
>Organization: Metropolitan State College of Denver
>Keywords: 199902181627.JAA08010 linux runlevel

Tony-

>I know this is hard to believe, but I think I've solved the mount problem
>on mclap.  BTW, editing the /etc/rc.d/rc.local file never did work.

What a unix geek you are becoming. ;-)

>The whole problem was that it was trying to mount wxbox before the network
>interface was started, so naturally it couldn't connect.  So, I started
>messing around with run level manager and found that if I added pcmcia at
>level 5 BEFORE network, portmap or nfsfs (the default is to add it AFTER
>nfsfs), it works !  Using the default position doesn't.

Well, I was going under the assumption that if the runlevel is set
to 5, then things in the previous runlevels (2,3,4) are done first.
pcmcia is started in 2, 3 and 4, so I figured by the time it got
to 5 to run nfsfs, the pcmcia interface should be up.  However,
it seems like on Linux, only the things that are in the runlevel
you are using are done.  I need to talk to our sysadmins about this
to confirm.  I talked to Mike about this before and he is the one
that led me to believe that the previous runlevels are done first.
That is the way it works on Solaris.  In fact, he advised against doing
what you did (changing the default position).  That was my first
thought as well.  But, if it works, I say go for it!

>I guess the reason the default settings work on winfo2 is that it isn't
>using pcmcia to start the network card, so copying the winfo2 settings
>wouldn't work on mclap.

Yup, that's right.

>Anyway, I'm feeling pretty proud of myself !!

Good sleuthing, Sherlock!  Who needs OS/2?

I'll let you know if I find out more from Mike or Sandy.  If you haven't
already, remove the mount command from rc.local.  No need to have that
done twice.

Lastly, we were on your system (wxbox) today and restarted the LDM
at about 10 AM.  We modified xcd_run to exec programs like ingebin.k, etc
instead of just running them.  The effect is to cut the number of
processes used for these programs in half (there were 2 for each program).
The ultimate goal is to figure out why ldmfail fails on your system.
We might get back to you in the near future to test our theories.

Don