perfSONAR User Q&A and Other Discussion

["Andrew Lake" <>]

Hi Carl,

Assuming it doesn’t go green, what happens if you set “filter_tool_name” to 0 in /opt/perfsonar_ps/mesh_config/etc/gui_agent_configuration.conf and then re-run /opt/perfsonar_ps/mesh_config/bin/generate_gui_configuration? It’s trying to only grab iperf3 tests (which is good) but I can’t tell if the tool name generated matchs what’s in your MA.

Thanks,

Andy

On Thu, Oct 8, 2015 at 8:23 AM, Garnizov, Ivan (RRZE) <> wrote:

Hi Carl,

Have no worries. It will go green soon, unless sudden failures appear. You have almost, but not quite, a full week of successful measurements. On the charts I looked.

Best regards,
Ivan

-----Original Message-----
From: [mailto:] On Behalf Of Carl Hayter
Sent: Mittwoch, 7. Oktober 2015 21:08
To:
Subject: [perfsonar-user] Maddash only working halfway

I'm having a bit of a weird issue with maddash. We have 8 perfsonar boxes (updated to 3.5) and I installed maddash one one of them, created a mesh configuration and configured all the machines to pull it down, created the maddash config, and crossed my fingers...

Funny thing is, on the dashboard, the OWAMP tests show up fine. The throughput tests all stay orange and say 'Unable to find any tests with data in the given time range ...'.

Looking at individual boxes, the tests in the mesh config are running and collecting data. If I go to hpc-perfsonar.usc.edu I can view the graph for the test against phe-perfsonar.usc.edu, and if I go to phe-perfsonar.usc.edu I can view the graph for the test against hpc-perfsonar.usc.edu.

On the maddash page, the HPC - PHE cells are orange and have the error message, if I click on the cells I do see a graph with the correct test data.

http://hpc-perfsonar.usc.edu/maddash-webui/index.cgi?dashboard=USC%20TEN-II

http://hpc-perfsonar.usc.edu/maddash-webui/details.cgi?uri=/maddash/grids/USC+TEN-II+-+USC+TEN-II+Throughput+Testing/phe-perfsonar.usc.edu/hpc-perfsonar.usc.edu/Throughput

http://hpc-perfsonar.usc.edu/maddash-webui/details.cgi?uri=/maddash/grids/USC+TEN-II+-+USC+TEN-II+Throughput+Testing/hpc-perfsonar.usc.edu/phe-perfsonar.usc.edu/Throughput+Alternate+MA

If I do a `curl http://phe-perfsonar.usc.edu/esmond/perfsonar/archive/`
I get a bunch of JSON data, same for hpc-perfsonar.usc.edu.

If I run the `check_throughput.pl` nagios script (I think this is how the maddash data is acquired...), I get seemingly normal data.

~~~ on hpc-perfsonar.usc.edu ~~~

H=hpc-perfsonar.usc.edu
P=phe-perfsonar.usc.edu
C='/opt/perfsonar_ps/nagios/bin/check_throughput.pl -p tcp --tool bwctl/iperf3 -w 7: -c 20: -r 86400'

# PHE->HPC from PHE ma
$C -s $P -d $H -a $P -u http://$P/esmond/perfsonar/archive
PS_CHECK_THROUGHPUT CRITICAL - Average throughput is 9.840Gbps | Count=9;; Min=9.60168;; Max=9.90942;; Average=9.84042555555556;; Standard_Deviation=0.121737880414458;;

# HPC->PHE from PHE ma
$C -s $H -d $P -a $P -u http://$P/esmond/perfsonar/archive
PS_CHECK_THROUGHPUT CRITICAL - Average throughput is 9.273Gbps | Count=7;; Min=5.64209;; Max=9.9051;; Average=9.27343285714286;; Standard_Deviation=1.60169745801878;;

# HPC->PHE from HPC ma
$C -s $H -d $P -a $H -u http://$H/esmond/perfsonar/archive
PS_CHECK_THROUGHPUT CRITICAL - Average throughput is 9.816Gbps | Count=4;; Min=9.76832;; Max=9.83687;; Average=9.8162625;; Standard_Deviation=0.0322087476473979;;

# PHE->HPC from HPC ma
$C -s $P -d $H -a $H -u http://$H/esmond/perfsonar/archive
PS_CHECK_THROUGHPUT CRITICAL - Average throughput is 9.542Gbps | Count=4;; Min=8.44071;; Max=9.9097;; Average=9.54223;; Standard_Deviation=0.734346699182354;;

~~~

So, how/why does my dashboard show the OWAMP results correctly but not the Throughput results?

I've attached the 'usc-tenii.conf' (mesh config file), 'usc-tenii.json'
(via /opt/perfsonar_ps/mesh_config/bin/generate_configuration), and 'maddash.yaml' (via /opt/perfsonar_ps_mesh_config/bin/generate_gui_configuration).

Any ideas?

----
Carl Hayter
ITS - Data Network Operations
University of Southern California