Panic

From Progzoo
Jump to: navigation, search

Things can go wrong at progzoo.net You may be able to fix them

Adding Amazon Servers

If the queue builds up we can buy CPU time at Amazon. They cost several cents per hour so please shut them down as soon as possible. http://tracker.napier.ac.uk/progzoo

Corruption at the server

This is mostly caused by a queue getting corrupted. Progzoo queues all requests so that there are only a limited number of jobs running at any one time.

  • The queue is stored in JSON format - due to Andrew's dumbness this can get corrupted.
  • In that case you can trash the queue.
    • This means that all waiting jobs will be lost so don't trash it unless
      • it is corrupt (because when it is corrupt no one is going to get served)
      • you are authorised to do this
      • you really, really want to

You can see the queue by visiting http://progzoo.net/r.cgi?lang=java Change lang to one of java cs cpp pl php py rb vb

You can trash the queue by visiting http://progzoo.net/r.cgi?lang=java&trashTheQueue=0 - but change the 0 to 1.

Before trashing the queue please copy it and send it to me at ajcumming@gmail.com with any other information that might be useful.

Restart the virtual server

The command /home/andrew/runUML.sh contains the following:

rm pz.fs.cow
screen -d -m /home/andrew/kernel64-2.6.24 ubda=/home/andrew/pz.fs.cow,/home/andrew/pz.fs mem=228M eth0=tuntap,tap1 umid=umid

Another way to kill the guest server(s) (if you don't have mconsole)

kill `cat .uml/*/pid`