[ltt-dev] LTTng net_extended_trace - example data
benjamin.poirier at polymtl.ca
Wed Feb 3 14:36:48 EST 2010
You can find a set of traces, results57, which work with the newest lttv at
These work with lttv 0.12.29
and with the current git head.
Compile and install lttv from the tar archive, then run:
lttv-gui --sync --sync-stats -t results57/node1 -t results57/node2
If you are confortable building from git, you can compile and then run:
./runlttv -a"--sync --sync-stats" results57/node*
These traces showcase two programs: "broadcaster" which is a traffic
generator and an ssh session, which is more interesting.
Have a look around time 3092s, pid 1896 on node2 is the ssh client and
pid 1927 on node1 is the sshd server. The two are running on different
machines but the traces align accurately. In this case, the average
accuracy is +/- 28us. You can compare this without synchronization:
rerun lttv without the --sync options. The client and server are miles
If you'd like to try out advanced synchronization options, have a look
at lttv/lttv/sync/README. If you encounter problems or have further
questions, you are welcome to contact me or the ltt-dev mailing list again.
paljak at mit.bme.hu wrote:
> Dear Benjamin,
> I am writing about your net_extended_trace extension to LTTng, I am
> evaluating it for a project. I looked up your e-mail announcing it on
> the lttng-dev mailing list, and I wanted to download the sample data
> from here: http://hal.drap.polymtl.ca/results55.tar.bz2 , but it is no
> longer available.
> Could you send it by e-mail or make it available somewhere online? I'd
> like to check out a sample before digging too deep into LTTng if it
> fits the needs of the project.
> Thank you very much in advance!
> Yours sincerely,
> Gergely Paljak
> PhD Student
> Budapest University of Technology and Economics
More information about the lttng-dev