We have upgraded our own c++ library and the system is working again now.
However, our monitoring daemon queries the CoMon
daemon for the disk usage on each PlanetLab node.
It seems CoMon currently cannot return the disk usage correctly.
So please use disk >= 0 for the queries. -- Jin (5/17/2007)
PlanetLab recently upgraded to 4.0 (OS and C libraries).
As a result our system is not working at the moment.
We will solve the issue as soon as possible. -- Jin (5/7/2007)
The firewall issue has been solved by moving the resource manager to
a planetlab node. The queries should work now. -- Jin (10/24/2006)
Recently our department has installed firewall rules that
prevent arbitrary planetlab nodes to report data to our
local nodes. So the query is not working right now.
We'll address this problem shortly. -- Jin (9/25/2006)
Specify requirement on CPU time, memory and
disk and find out the planetlab nodes that satisfy the requirements.
You can also download a library API and the source code of
a query client in this package.
The source code shows how to easily integrate the query
capability into your C++ programs.
Some explanations:
- Available CPU time means the percentage of CPU time that
a new process is entitled to. It is computed as
1/(current_load_average + 1).
- If many nodes satisfy the resource requirements, some
random entries are returned. Thus two queries with
exactly the same condition will most likely
get different results.
- If the resource requirement is too high, fewer or no
entries will be returned.
|
Please direct any questions to |
 |