[Planetlab-users] Collecting log-files

[Emil Sit]

On Tue, 24 April 2007 at 12:38 (-0400), Chad Yoshikawa wrote:
> In my experience the CPU is scarcer than the network bandwidth, so I'd go
> with a time-staggered remote copy of the logfiles back to your machine.

Pulling the files locally may make it easier to re-process your logs
when you realize that there's a bug in your post-processing scripts,
or something you forgot to extract.  Were you going to delete the
raw log files from the PL nodes after copying them?

> > A: process the logfiles locally on each planetlab-node (run it through a
> > grep+sed-script) such that the amount of data transmitted is minimized

On the other hand, grep and sed usually aren't too CPU intensive;
I'd imagine you'll more likely be bottlenecked by the speed at which
you can pull the files off the disk. 

Incidentally, do you need both grep and sed?  (You can do much of the
same stuff in just sed.)

-- 
Emil Sit / http://www.emilsit.net/