[Planetlab-users] Nodes Wedged?

[Neil Spring]

On Jul 31, 2007, at 2:43 PM, Bryant, Jason R Ctr AFRL/IFSB wrote:

> Working with the cornell nodes on planetlab I have had several  
> start to
> refuse access to me for the last several months.
> Planetlab2.cs.cornell.edu, planetlab3-dsl.cs.cornell.edu, and
> planetlab5.cs.cornell.edu allow me to start the login process, but  
> then
> either give an "Device or Resource busy" error, or an "Remote host
> closed connection" error.  The cornell admin suggested removing and  
> then
> adding the nodes to my slice again, so that the user list would be in
> sync, and to try again several hours later.  I tried this but the same
> errors continue.

If you see an error like:

src/vsh.c:309: pl_chcontext(702): Device or resource busy (16)

That's the time to remove and re-add that node to your slice.

Other times, the node is likely just down or otherwise broken (comon  
can help figure out what went wrong and when).  If you have a  
relationship with the cornell admin's, their help might get the  
machines back on line more quickly.  They may have to update the boot  
cd or manually reboot the machines that you notice have failed.

PL staff have been working a lot on the node manager software whose  
bug causes this error, so it might soon be obsolete.

-neil