Experimental Physics and
Industrial Control System

[email protected] (Jeff Hill) · Thu, 18 Nov 1999 19:43:02 -0700

Nick,

I must admit that we recently saw a very similar situation
here in one of our systems, and we are still attempting to track
it down.

We are currently attempting to isolate between:
o the CA server code
o the DB CA link code

I have been looking carefully at the CA server and I don't see it. Bob
has also been looking at the DB CA link code.

> I reset one of the link fields in the record doing the monitoring (it was
> a Calcout record and had many CA input links from more than one other
> machine) by writing the the current value of the link to it. This seemed
> to free things up and all the CA monitored values in that record got set
> to the current values.

If you poked, on the server side, the input that had a different value than
what was expected on the client side then it could be either CA or the DB
CA link code. If you poked, on the server side, some other input whose
state was properly reflected on the client side and things got better then
I would expect the DB CA link code (or possibly the CA client library) is
at fault.

> I finally rebooted the IOC to clear everything up.
>
what was the functionality that could not be restored without a reboot?

>
> I am running R3.13.0Beta11. The machine doing the monitoring is a Baja47
> MIPS (RISC) CPU. The other machines are Baja60's (68060).

Since it appears that in this situation the servers are not running
on RISC machines then I don't see an immediate tie in with the problem that
Noburo discovered.

I will look at this in further detail tomorrow.

If anyone else has seen a problem similar to this I am very interested in
hearing details about the surrounding circumstances (like the client
side program that was running).

Jeff

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System