| Title: | FDDI - The Next Generation |
| Moderator: | NETCAD::STEFANI |
| Created: | Thu Apr 27 1989 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2259 |
| Total number of notes: | 8590 |
Query from customer to FE...comments?
-------------------------------------------------------------------------------
The Gigaswitch card in question had FSASK2 connected to one port and a
Xyplex SX-2550 FDDI concentrator connected to the other. Both are SAS
connections. The FDDI concentrator had FSASK3 and FSASK4 attached originally.
Gigaswitch
Card
+---+
| |
| |-|---- FSASK2
| |
| |
| |
| |-|-
| | \
| | \
+---+ \ FSASK3 FSASK4
\ \ /
+-\--------\-----/-----------------+
| \ | | |
| -- -- -- -- -- -- -- -- |
| |
+----------------------------------+
Xyplex SX-2550 8 Port FDDI concentrator
Kathy and I then configured FSASK9 into the concentrator. We them saw a
multitude of problems:
- FSASK2 would have cluster connect timeouts
- FSASK4 would crash
- FSASK9 would drop sessions
The first thing we did was remove FSASK9 and FSASK4 from the concentrator.
That stablized the cluster. I them noticed a high number of the Ring
initializations initiated, Ring initializations received, Ring purge errors,
and FCI strip errors on both FSASK2 and FSASK3. I thought that they may be due
to the two connections that FSASK3 and FSASK4 had on the network so I took the
duplicate connections off of the network. I still saw the errors but not right
away. It took some large file copies before the errors showed up again.
Traffic seems to make the problem more pronounced.
Today I removed the concentrator from the network and put FSASK3 directly
into the Gigaswitch. I have not been able to reproduce the errors, even with
some large file transfers going through the card.
There are still a number of things that can be wrong. The concentrator
will have to be connected back onto the network to eliminate some of them.
1) A bad cable between the concentrator and the Gigaswitch.
There will be a new cable between the concentrator and the Gigaswitch
when the concentrator is put back onto the network.
2) A bad card in FSASK3.
When FSASK2 started dropping out of the cluster, the FDDI card was
replaced. It was not replaced in FSASK3. The reboot may have cleared
the problem up temporarily so I will leave the concentrator off until
at least Tuesday night to see if the errors returned.
3) A bad concentrator.
A possibility. The concentrator is a fairly passive device so there
is not much in the way of errors in the counters. What concerns me
more is that either the Gigaswitch cannot handle a concentrator in the
port or there is some compatibility problem with the Gigaswitch and a
Xyplex concentrator.
If there is a problem with concentrators in a Gigaswitch port, you will
have to find that that out for me. You may also be able to find out if
there have been problems with Xyplex concentrators and Gigaswitches
working together.
Another concern of mine is that the Frame Content Independent (FCI)
Stripping and Ring Purger algorithms used by Digital may be causing
compatibility problems with the concentrator. Digital developed their
own algorithms and submitted them to the ANSI FDDI Committee. Part of
the submission was used to update the MAC standard. I am not sure how
standard the algorithm is though. However, as I said before the
concentrator is a passive device so this is likely not an issue.
This question could be solved by getting another concentrator from
Xyplex and one from Digital. If the new Xyplex concentrator worked, it
was a bad concentrator. If the DEC concentrator worked and the new
Xyplex one didn't, it's a compatibility issue. Wayne, what are the
odds of grabbing an FDDI concentrator from DEC to try?
(Note that if it was a compatibility issue, I would rather replace the
Gigaswitch than the concentrator. We definitely need standards to be
followed by all our networking vendors.)
All in all, one other thing concerns me. Each of these problems were
isolated to the one port on the Gigaswitch. It was my understanding that each
port acted as a separate FDDI ring. If so, why were both ports affected? More
to the point, a large network file copy from FSASK2 would cause errors on
FSASK3 - FSASK3 was not doing any copies.
The only action item for you right now Wayne is seeing if you can grab a
Digital FDDI concentrator. I will keep you informed on the error counts on
FSASK2 and FSASK3.
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1417.1 | NPSS::MDLYONS | Michael D. Lyons - Young enough and dumb enough | Mon Aug 08 1994 18:39 | 22 | |
I know of no known interoperability problems with the GIGAswitch
and Xyplex concentrators.
Please provide some more details about the configuration,
specifically, what are the interfaces in question, what are the
revision levels of the firmware/hardware involved, including both the
GIGAswitch and interfaces. Do the GIGAswitch counters reflect the same
information as you reported on the system? I'd also suggest examining
the GIGAswitch error logs to see if there's any evidence there.
I'd strongly suggest you get a *good* FDDI analyzer hooked up and
see what is really going on.
...as for replacing the GIGAswitch if there was a compatability
problem, why? If there were an interoperability problem, I'd expect
the vendor causing the problem to fix their equipment, and
expeditiously. (This includes both us and Xyplex.) If this were an
interoperability problem caused by the Xyplex, dumping the GIGAswitch
would only postpone the problem until they connected another vendors
equipment.
MDL
| |||||
| 1417.2 | CSC32::J_SHUMWAY | Mon Aug 08 1994 19:07 | 2 | ||
I too found the attitude "get rid of the Gigaswitch" rather
interesting. Waiting on info from the field...
| |||||