perfSONAR User Q&A and Other Discussion

[Kate Adams <>]

Hello All,

Thank you for your response, Andy.  I changed force_bidirectional 1 just to see what would happen, I'll change it back now that you've answered that part of the question.

Running /usr/lib/perfsonar/scripts/system_environment/configure_esmond --force did not change the key.

Is there a particular error in a particular log file I should be looking for to see why tests can't archive?

On Fri, Oct 5, 2018 at 11:13 AM Andrew Lake <> wrote:

Hi,

force_bidirectional is a very confusing option in my opinion, which is part of the reason we started recommending against it in 4.1. Before I get too much into this specific case a much more detailed look at force_bidirectional and how it relates to dashboards see http://docs.perfsonar.net/psconfig_meshconfig_migrate.html#special-notes-on-force-bidirectional-and-changes-to-maddash-display

As for your dashboards, the common denominator in the all orange GPN column in the first grid and the all orange GPN row in the second is both are trying to store their results in the archive http://speedy.greatplains.net/esmond/perfsonar/archive/. It would appear it is having trouble doing so. Based on other emails on the list, I'd start by running "/usr/lib/perfsonar/scripts/system_environment/configure_esmond --force" and take note if it changes the "_auth-token" field in /etc/perfsonar/psconfig/archives.d/esmond_local.json. The initial migration scripts we released could allow for that value to get set incorrectly.

Thanks,
Andy

On Thu, Oct 4, 2018 at 4:39 PM Kate Adams <> wrote:

Hello All,

Has anyone else experienced this curious behavior on the maddash for version v4.1.2-1.el7?

Here is a link for reference: http://ps-dashboard.greatplains.net/maddash-webui/index.cgi?dashboard=GPN%20Mesh

This email is about the "Great Plains Network" horizontal and vertical lines in the graphs, but there are other nodes with the same behavior.  However, not all behave like this, hence this email.

force_bidirectional 1 is set so there are two graphs for every test. This make sense.

On one set of graphs, say traceroute since it's the first pair of tests, GPN is orange on the vertical, but pretty green on the horizontal.

On the second graph, GPN is pretty green on the vertical, but orange on the horizontal.

Is this expected?  If this is a result of force_bidirectional 1 (like the two graphs are), why don't both horizontal and vertical lines show up as green when force_bidirectional 0 is set?

Thanks in advance for any insights,

Kate

--

Kate Adams ()

Cyberinfrastructure Technologist

Great Plains Network

--

Kate Adams ()

Cyberinfrastructure Technologist

Great Plains Network