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

20010214: packet sniffing sheds a little light (cont.)



>From: Unidata Support <address@hidden>
>Organization: UNCA
>Keywords: 200102052335.f15NZkX27276 McIDAS-XCD

Leigh,

So, what I have found after beating my head up against the problem
of serving sounding data off a remote ADDE server hosted on Linux
is that the problem apparently appeared with RedHat 6.x (I have
verified 6.2, but I don't have access to 6.0 or 6.1 machines).
The same McIDAS code on a RH 5.2 machine runs _without_ error!

We are attempting to isolate exactly where the cause of the problem
lies.  We have tried dropping back to the version of gcc that was
distributed with RH 5.2, but this does not see to make a difference
(but the executables created are using the 6.2 shared libraries).
The next test will be to try and get things running using the
shared libraries from 5.2 to see if that makes any difference.

One thing I have noticed is that every machine that exhibits the
failure is running a SMP Linux kernel.  Do you have a single
processor machine at UNCA that we could get the remote ADDE server
running on and test out if the problem is somehow related to
the SMP kernel?  I am grasping at straws here...

This one is a real bugger!

Tom

>From address@hidden Wed Feb 14 21:00:55 2001
>Subject: Re: 20010214: packet sniffing sheds a little light (cont.) 

Interesting... could be a glibc 2.1 vs 2.2 thing.
I know a lot of things broke when that switchover occurred... I think it
occurred between RH5 and RH6.

storm2 is a uniprocessor machine and it is running the single-processor
version of Linux... so perhaps it's not a SMP thing.

Leigh