Friday, November 25, 2011

When something goes wrong ...

... it usually happens in more than one place at a time.

The laptop I use as train computer developed a memory problem. The BIOS disables any memory above 88M due to a failure during POST. Yes, that laptop was quite slow, but it did the job, and has a beautiful, huge 1900x1600 display. Evaluating my options:

  1. I can go out and buy a cheap desktop computer, put it in the server rack, and run VGA/keyboard through the KVM to the 15" monitor at the rack. That would likely work out to be a DualCore or low-end i3.
  2. Ditto, but get just a low-end, low-power motherboard, and memory. One of the current pseudo-embedded CPUs, or Atoms, should do just fine. Install it in a spare computer case and put it into the rack.
  3. Try and make do with the 88M memory, convert the laptop to a dumb X-Terminal, (potentially remote boot it), and run JMRI on the server with a remote X session to the laptop. Technically a bit more interesting, a bit more involved, and actually not really something I'm particularly keen on. I get enough overly complicated setups at work.
I think I'm going to throw hardware at the problem, and go with either option 1. or option 2. and give up on the nice display.


The Emsingen power district of the layout developed a short circuit some time in the last 48 hours and I have no idea what caused this. The PSX circuit breaker is solidly reporting a short. All 3 PSX's LEDs flicker roughly  every 2-3 seconds, and the amp-meter shows an inrush of current at the same time. Evaluating my options:
  1. When I installed semaphores yesterday, something got on the track and is causing a short. I brushed down all tracks in Emsingen. No dice.
  2. Maybe a locomotive has a shorted out decoder, or some car is derailed. Removed all rolling stock from Emsingen. No dice.
  3. When I cut and glued down the north stub track next to the loading ramp on Wednesday, I somehow caused a short. Unlikely, but I ripped out this piece of flextrack anyways. No dice... Now that's annoying.
  4. I somehow jumbled track wires underneath Emsingen. No idea how that even could have happened. Did a visual check, and all connections look good.
  5. Powered off everything. Maybe the PSX just got confused. Nope, problem still there.
  6. Maybe the PSX is bad. Put the district on another PSX for testing. Still to do.

Patricia's computer developed an interesting file system problem. It appears that when the system stats some inode linked from her home directory, the operation hangs forever. This causes process like Nautilus or ls to go into state D+. I can kill -9 the process from another terminal, and get back the prompt. I can do tab completion (e.g. type ls .[TAB] and get a list of all files starting with '.'). I haven't tried every character yet. There are no hard disk related errors in dmesg. smartctl reports the hard disk recorded no internal errors either. I'm suspecting ext4 file system corruption, and will try an fsck next.


Bottom-line: What the hell is going on today?


Bernhard said...

Wild... The file system problem disappeared after a reboot (and a bunch of debugging that went nowhere). Even fsck couldn't find anything wrong with the file system. Kernel choked on something? This is One would think they have stuff like this worked out by now.

Bernhard said...

And just in case you where wondering why our Web site didn't work since yesterday ... Yes, I accidentally pulled the power plug from the switch in the server rack ...

Bernhard said...

It's not the PSX. When putting the Emsingen district on another PSX, the short indication moves with the district.