How Monitor Works

Our goal with Site Assistant and Monitor is just to keep the machines registered with PlanetLab up and running PlanetLab software. This gives the researchers who use PlanetLab access to the resources at sites across the world.

One of the first steps we're taking to help with this is the Site Assistant service, called Monitor. As it runs, it follows a specific set of policies. These policies are explained in detail on the following pages, but in summary, the basic ideas are below.

Please note that things will escallate from week to week only if no one responds. We will work with you. These are the only steps we can take if there's no communication.

Sites

  • Week one, email the Technical Contact about the problem we see and suggest how to correct it.
  • Week two, email a similar message to the Technical Contact and PI, and disable slice creation.
  • Week three, email the Technical Contact, PI, and Slice users, and disable all slices on PlanetLab.
  • Ever after, send periodic emails in case anyone is listening...

 

Site Policy

 

Each member site of PlanetLab receives some Slices in exchange for two machines running the PlanetLab node software. For all sorts of reasons, the machine can stop: maybe due to lost power, kernel crashes, or network disconnection. Who knows. To follow what's going on, Monitor periodically looks at the status of the machines at your site and makes a note of it.

After a machine is down for longer than one week, Monitor will send a message to the Technical Contact at your site. System administrators are busy, busy people. We totally get that. And their first priority may not be responding to a maintenance request for a PlanetLab machine.

However, if we don't hear back from someone acknowledging that they'll look into the problem and the total number of running machines at your site drops below two for longer than two weeks, then we will contact the PI at your site and slice creation is turned off. As a result, no new slices can be created, but the existing slices will continue running on PlanetLab uninterrupted.

If we still don't hear back from either the PI or Technical Contact after three weeks, then all slice users will be emailed and all slices on PlanetLab will be disabled. The idea is that anyone actually benefitting from PlanetLab will have some added incentive to get the machines back up and running. If we continue to not hear from anyone, then periodic emails will be sent after this to remind everyone about the situation.