fcmsutil excessive Bad Rx Char

[Steven M Evans]

We have 10 Tachyon XL2 cards in an rp8400. 5 are in use, 5 are not. We had a problem with one of the ones in use, and we think it's the cable. While troubleshooting that, I noticed the 5 that are not in use have excessive "Bad Rx Char" and "Loss of signal", like so:

Loss of signal 1520 Bad Rx Char 2325600
in 2 hours.

Out of the 5 not in use, I plugged in a loopback cable, and that did nothing. The other 4 have cables leading to nowhere.

What might be causing this, and how do I fix?

[Steven E. Protter]

Shalom,

I suspect a bad cable at this point. I doubt two cards are down at the same time.

It is also possible, you misdiagnosed the troublesome card and switched the cabling from the wrong one.

Could also be bad switch port or whatever the other end of the cable is connected to.

SEP

[Florian Heigl (new acc)]

Some conspiracy theories about the excessive errors on the unused cards:

- A FC AutoNeg configuration mismatch (As You can't set it manually on HP-UX, a look at the remote end might be more interesting
- The switch port is disabled via software but still shows some light on Your RX port

- If there really is nothing connected to the remote end, then I'm stunned - usually the system should be quiet and happy. A quick fix would be to use /opt/fcms/bin/tdutil /dev/fcN disable

I'd probably start by letting HP check the HBA's firmware revision.

[Steven M Evans]

SEP -
There was a bad cable on one of the five 'working' paths. We positively ID'd the bad cable. The five 'not in use' are the ones I'm asking about. There is nothing on the other end of 4 of these cables, and a loopback on the end of one; all 5 complain.

Florian -
I think your on the right track. The power light (left) is on, and the link light (right) is off on all 5 cards. I know for certain that 2 have nothing on other end, so I don't think AutoNeg is a problem.

What do you mean by "the switch port is disabled via software"?

Disabling the card does keep it quiet, but I think that is just a workaround. Is there a way to keep this persistent across boots?

Before posting, HP told me to remove or unseat the cards - I don't think I was talking to the right person, and getting navigated to the right person at HP is sometimes.... I will ask about firmware revision.

- SME

[Florian Heigl (new acc)]

Sorry this reply is coming real late now.

if there is nothing at the end of the cables, then You can ignore my question about a switch.
if the 'nothing' includes a SAN switch, we need to talk again.

Two more tests:
do the errors go away or decline in rate
when You
- put the dust covers at the cables end?
- disconnect the cable from an hba?

It's weird Your adapters think they have link when they shouldn't, and I definitely haven't seen that yet. At our site, systems will sometimes be disconnected from the SAN, but usually at the host end, and I don't recall seeing excessive linkside errors by that.

About Your HP case I don't have good advice, we have always received very good and able support, but there might be a relation to the customer's size and contract.
So I'd just recommend You update the case via email with a lot of additional data and some explanations.

Still firmware is my #1 try, I'll try to figure out how to query the firmware revision online and check what our test boxes' XL2 adapter has (it has been unplugged quite a few times ;>