[Planetlab-users] PlanetLab Status

[Scott K. <sc...@CS...>]

PlanetLab is currently experiencing trouble on a couple of fronts.
The implementation team is working diligently to get things running
smoothly as soon as possible.  This message is to provide everyone
status and, as a consequence, reduce the flow the tide of duplicate
messages to planetlab-support.

Please feel free to forward this information within your organization
as not everyone is subscribed to planetlab-users.  This is also a
request for everyone to subscribe to planetlab-users and planetlab-announce.

www.planet-lab.org
------------------
  * This week, our primary web server failed.  The problem was
    exacerbated by the winter storm in Hillsboro, Oregon, preventing
    the west coast implementation team from getting to the machine.
  * In the past day, we have remotely brought up an alternate server.
    Most (but not all) of the web content has moved; in addition, we
    are still in the process of moving the mail server.  Until this
    happens, e-mail notifications (e.g., when you request a password
    reset) will not work.

Kernel Bug
----------
  * Version 2.0 of the PlanetLab software appears to trigger
    a kernel bug related to semaphores leading to hung nodes.
    Prior to Version 2.0, we were seeing this behavior occasionally;
    it appears that we are now tickling this bug much more often.
  * This occurs on a relatively large set of nodes (approximately 30%).
  * We are currently testing more recent patches to the kernel
    and to the vserver subsystem.

Key Distribution
----------------
  * Additionally, we are improving the key distribution so that
    keys that are updated on the web are properly propagated to
    nodes that are already assigned to slices.  Currently, if you
    update your key, it will only be propagated to new slices
    and to nodes added to existing slices.
  * We expect to deploy a fix in 1-2 days.

Scott Karlin