[chulofiasco]

So I managed to get my hands on the Internal Support Tools, version 2.7 and in addition to Odyssey Diags, there is also an enhanced IDE testing environment for a few different models of workstation. I installed it onto my Octane for some rounds of troubleshooting a random kernal panic/memory fault. While banging around in IDE, it got around to telling me that my IP30 was defective, basically because it failed one of the DUART tests (which if I'm reading the screen right, is more than likely Serial Port B Test.)

Of course, immediately after beginning and not too long before it fails, it reminds me to "Ensure that Serial Port Plug is connected". With nothing connected, it then encounters an error, and reports that the PCI Serial Port Test failed. ErrCode PCI007. Of course it looks like a false positive because ** I think ** it might need this plug to test successfully, so I'm not super concerned, but then again, who really knows?

Is anyone familiar with this, or know what a "Serial Port Plug" is?

I've added some photos.

Thanks,
Lorenzo

[vishnu]

So I managed to get my hands on the Internal Support Tools, version 2.7

Wow, that's a rare beast, it was only supposed to be used by Silicon Graphics personnel wherever did you find it?

SGI error messages were deliberately obfuscated to prompt/motivate customers to call an SGI field service engineer to come out and fix things. Or, more particularly, to buy a service contract. The highest level contract guaranteed an on site response anywhere in the world within 24 hours. SGI made a lot of money from those contracts...

[chulofiasco]

So I managed to get my hands on the Internal Support Tools, version 2.7

Wow, that's a rare beast, it was only supposed to be used by Silicon Graphics personnel wherever did you find it?

I came across a very knowledgeable individual who was very kind in suppling this... Lots of interesting things on the disc, none of which tells me about Serial Port Plug or B Test.

SGI error messages were deliberately obfuscated to prompt/motivate customers to call an SGI field service engineer to come out and fix things. Or, more particularly, to buy a service contract. The highest level contract guaranteed an on site response anywhere in the world within 24 hours. SGI made a lot of money from those contracts...

Seems to be a legit observation. I'm not sure that the error is something super complicated, if it is really crying about a missing diagnostic aid. Without another Octane to test the theory (that it too will fail the same test due to a missing plug) I can't say for certain, but it could be a false positive. I tried to build a loopback plug, but that didn't really get me too far...

[vishnu]

Okay, so this is little more than wild Irishman hypothesizing but IRIX, at least at some point, supported TCP/IP connections to Ethernet using serial port to RJ45 converter cables, so, if you have one of those or can find one you could plug it from your workstation to your hub/switch and give it another try.

[jpstewart]

The "Serial Port Plug" in question is likely just a loopback plug that connects the machine's output to its own input.  They're a common way of testing the serial port:  write a bit pattern and if you read back the same pattern, then you know the serial port can transmit and receive properly.

Add the word loopback to the phrase and Google for it.  You can buy the plugs for $5 or build your own.

[chulofiasco]

The "Serial Port Plug" in question is likely just a loopback plug that connects the machine's output to its own input.  They're a common way of testing the serial port:  write a bit pattern and if you read back the same pattern, then you know the serial port can transmit and receive properly.

Add the word loopback to the phrase and Google for it.  You can buy the plugs for $5 or build your own.

Thanks, guess you missed above where I said I attempted to build a loopback but didn’t get very far.  I’ll go back and check again though.  🤷🏼‍♂️