perfSONAR User Q&A and Other Discussion

[Daniel Spisak <>]

Thanks for that piece of info, it was helpful. I was able to get a throughput test to run, but not the latency (however since this will all run in VMs Im not planning to do one-way latency anyways) and the "--sliprand" command line option is not valid. 

Do you or anyone have an example of a maddash yaml that works with 4.0?

Additionally, does anyone have any documentation on pscheduler as that seems to be what I should be fiddling with.

On Tue, Feb 21, 2017 at 4:42 AM, Antoine Delvaux <> wrote:

Hello Daniel,

With your 4.0 RC setup you should validate the testpoints using pscheduler rather than bwctl.  bwctl is only kept for backward compatibility in the 4.0 release and is not started by default (and shouldn't be).  I'm not sure why it is listening on an IPv6 socket though.  I'd suggest you try pscheduler with commands similar to:

pscheduler task --sliprand latency --source 172.28.128.xxx --dest 172.28.128.4

pscheduler task --sliprand throughput --source 172.28.128.xxx --dest 172.28.128.4

For the MaDDash configuration, you need to edit maddash.yaml file and remove the "My Sites" entry that is defined in the dashboards list.  Then restarting maddash should remove it from your menu.

Cheers,

Antoine.

Le 21 févr. 2017 à 03:43, Daniel Spisak <> a écrit :

Hello there! Been making slow progress on my automation here, have managed to get maddash to show my test mesh but I have run into more problems. Most of this stems from its not clear to me how to validate if a testpoint is functional or not. I've been trying to do manual bwctl commands from one virtualbox VM to another (they are using a common internal network and can ping each other). I can verify that both hosts have bwctl running and I've gotten ntpd synced up. I also made sure the internal network /24 was allowed in the bwctl-server.conf file as well. I also tried turning off the iptables rules as well. However when I try:

[root@testpoint1-centos-71 vagrant]# bwctl -T nuttcp -f m -a 3 -t 10 -i 1 -c 172.28.128.4

bwctl: Unable to connect to 172.28.128.4

[root@testpoint1-centos-71 vagrant]# bwctl -T iperf3 -a 3 -f m -t 10 -i 1 -c 172.28.128.4

bwctl: Unable to connect to 172.28.128.4

then, checking on the second testpoint VM that is the target:

[root@testpoint2-centos-71 bwctl-server]# netstat -tulpn | grep bwctl

tcp6       0      0 :::4823                 :::*                    LISTEN      3442/bwctld

I tried putting the ipv4_only directive in the respective test_specs for the central config file and regenerated the json file but it seems like bwctl is still listening only on tcp6 sockets and not tcp. How do I fix this? As long as bwctl listens only on a ipv6 socket my VMs will be unable to do any tests it seems.

Additionally, if someone could please tell me how to get rid of the "My Sites" listing under Dashboards for MadDash that would be great. When looking at the test mesh I created in Maddash all two nodes show as unable to retrieve data and when clicking on the link for the problems found I get:

Category:CONFIGURATION
Potential Solutions:
Check your maddash configuration

Which is not exactly the most helpful of errors. I have included my meshconfig-guiagent.conf below. Bear in mind that I am doing all the automation work and testing in a multi-VM virtualbox environment along with test kitchen, this is why you see RFC 1918 IP space being used. In production this will all use real routed IPv4 space.:

<mesh>

    configuration_url             http://172.28.128.5/perfsonar-mesh.json

</mesh>

maddash_yaml                 /etc/maddash/maddash-server/maddash.yaml

check_interval 3600

check_config_interval 60

send_error_emails_to_mesh     1

# The default maddash test configurations. If the values aren't specified in

# the mesh, these will be used.

<maddash_options>

    <perfsonarbuoy/bwctl>

        check_command            /usr/lib64/nagios/plugins/check_throughput.pl

        check_interval           14400

        check_time_range         86400

        acceptable_throughput    900

        critical_throughput      500

    </bwctl>

</maddash_options>

I have also tried regenerating the maddash.yaml file via restarting the meshconfig gui agent on the centralmanagement host which has the central config and that hasnt seemed to fix any problems either.

Help me Obi-Wan Kenobe you're my only hope!

Additionally, if there is some place on IRC where I can ask questions of people without having to clutter up the mailing list I'd be interested to know if such a thing exists. Thanks!

--

Daniel Spisak | Senior Dev Ops Engineer, Synthetic
AppDynamics
Monitor your Apps in Production. Now.
Watch our Video | Download AppDynamics for Free
Twitter | Facebook | appdynamics.com

--

Daniel Spisak | Senior Dev Ops Engineer, Synthetic
AppDynamics
Monitor your Apps in Production. Now.
Watch our Video | Download AppDynamics for Free
Twitter | Facebook | appdynamics.com