Today I went over to Brian's to work on some problems with the dilab. The remaineder of this blog entry is unlikely to be interesting to anyone except possibly sparc and mips porters, but I wanted to write it all down.

First thing I noticed was that the internal network was very slow and lossey. I isolated this to its switch, which must be broken. Brian lent me a 7 port hub, which I put the more demanding machines on, leaving the others on the switch, and hopefully this will clear up some of the network issues I've been seeing during tests, until I can install a proper new switch.

Unfortunatly, this still didn't clear up the proliant's failure to use its emulated network CD drives, which I'd hoped it would. Still working on figuring out why that's broken.

As I was testing out all the machines after that to make sure they still worked, I heard a very strange sound from one of the X-10 appliance modules (#3). Sounded like a CD drive failing to open. I think its relay must be fried, after 1.5 years of turning on and off half a dozen times a day. So I had to move the raq 1 on that over to a different power module (#1), and moved the ia64 that was on that one over to a non-controlled power port.

The raq 2, that handles d-i daily mipsel builds, was hung with "booting Debian" on its LCD, and power cycling didn't help, unfortunatly it lacks a serial console to see what's wrong. So I pulled it and took it home, will have to investigate it and try to get a stable system on it. Until then mipsel d-i dailys are still down, until ths takes them over.

Finally, I took a look at the 2.6 kernel issue on the sparc ultra 5, which worked with 2.4 but not 2.6. Luckily Brian knows a lot about suns and a lot about serial (used to be my sysop, and still is in a way), and he diagnosed it as having the port running too fast, dropping it down to 9600 fixed that problem.

So, shopping/todo list from the day:

  • I need 1-3 more X-10 appliance modules or tranceivers.
  • I need a new 16 port switch. That'll be the third one, the lab's already eaten two now.
  • I need at least one more serial null modem cable for the raq 2.
  • Need to diagnose the raq 2.
  • Still need to figure out why ILO on the proliant doesn't like my web server's CD images anymore.
  • The a-500 is still not seeing its PCI bus.
  • Need to rewire the whole rack with better patch cables and also moving all these wires around has created a bit of a mess.