Early this week I upgraded my workstation [DEC 3000 Model 700] to DU
4.0. Actually, I did a full install. Now, I have two rather serious
problems, and although I have a software maintenance contract with
Digital, they don't return my calls. So, I ask here if anyone knows
about these:
An important piece of background is that my workstation is connected
via an ATM network. I am using the Digital ATMWorks 750
turbochannel card. Our ATM switch is a high end FORE ASX.
Problem 1:
When I initiate network activity to DU 3.2[a,c,d] machines,
the SVC never comes up. The arp'ing completes, however. Even worse
is a very high probability that the DU 3.2 machine will kernel panic
due to an illegal memory access. This gives new meaning to "killer
workstation." If the DU 3.2 machine initiates the contact, then the
circuit comes up and all is well. The DU 4.0 machine can succesfully
initiate contact with Solaris 2.5 and HP-UX 9 hosts.
Problem 2:
I get lots of RPC timeouts when writing via NFS2 and NFS3. The
machines I've tried are all on the other side of a BayNetworks router.
I connect to the Baynetworks router via a PVC. NFS reads work fine.
I've set my MTU to 1500 on the ATM interface, and the NFS servers also
have an MTU of 1500 on their FDDI and Ethernet ports respectivly.
I've tried turning off rpc.statd and rpc.lockd on my local workstation,
but that makes no difference. Is it possible that the DU4.0 lis0 interface
ignores the MTU setting? This is exactly the sort of trouble I've had
before when the MTU of the client was much larger than the MTU of the
server and they were seperated by a router.
Thanks,
-Henry
Received on Sun Jun 23 1996 - 05:08:25 NZST