We think we know what's happening. This message was introduced to VM
in an apar that added Ficon functionality. (I found this on
Google.) We don't use Ficon on the mainframe, but we do use it on
the "other side" (the dark side) of the Shark, for some new Blade
servers. They've been doing experimenting over there to see what
happens when they disconnect various things. The hypothesis at this
point is that the Shark is reporting the incident to the mainframe
side even though it originates with the Ficon on the open systems side.
I suspect the "recoverable machine check" in EREP is just a vehicle
used by VM to record such miscellany.
We'll do some testing to assure ourselves this is what's
happening. Undoubtedly our CE will deal with the Shark support people on this.
Thanks for your suggestions. I wouldn't have posted the question if
we had figured all this out, but that light bulb didn't go on until
later last night.
- Tom.
At 05:11 PM 8/2/2006, you wrote:
> We have a Shark, and we've been getting HCPA2F2234E LINK FAILURE
> messages with increasing frequency. The messages point to the Shark
> by model and serial number. The message text says:
>
> "A link-maintenance information record has been written to
> LOGREC in response to the reporting of a serial incident."
>
> VM EREP shows recoverable machine check records corresponding in
> date/time to these messages.
> [snip]
> Anyone have any idea what these messages are prompted by and what
> they're trying to tell us?
Have someone experienced with the use of an optical reflectometer check
the connectors on the cables. A loose or imperfectly formed joint on a
fiber jumper will cause symptoms like this.
Another possibility is a marginal laser diode in one of the adapters. If
the diode output waveform is skewed beyond a certain threshold, you'll
get a recoverable machine check like this because redriving the I/O hits
in a randomly different part of the waver and everything works that
time.
In either case, you need optical diagnostics equipment to find this one.
What exactly is the EREP error coding and the log record dump?
-- db
Tom Cluster
County of Sonoma
Santa Rosa, CA
(707) 565-3384 (Tuesdays and Wednesdays only)
