perfSONAR User Q&A and Other Discussion

["Christopher J. Tengi" <>]

Andy,
        Thanks for the quick reply.  I suspect that my best bet is to go with 
“temporary” mesh configurations with “no_agent 1” directives.

/Chris


> On Dec 8, 2015, at 2:13 PM, Andrew Lake 
> <>
>  wrote:
> 
> Hi,
> 
> It sounds like you found most of your own answers :) Yes you can manually 
> add tests to maddash.yaml and the mesh will preserve them. Also, adding 
> mutiple mesh blocks to gui_agent_configuration.conf is the correct way to 
> point at multiple meshes. For the short to medium term tests it all depends 
> on personal preference as to whether you put them in a mesh config or do 
> them by hand. Some advantages of doing a meshconfig might be that you can 
> also point the testers at them and have them pick-up the tests. It will 
> also be easier to remove them since you won’t have to pick through the 
> maddash.yaml file and delete stuff. Generally I only add meshes by hand if 
> I am trying to do something with MaDDash that the MeshConfig doesn’t 
> support (e.g. using a check other than the standard bwctl and owamp one, 
> setting up a really sparse test topology). That’s just a personal rule of 
> thumb and you can add whatever you want in there. 
> 
> Also, https://github.com/esnet/esnet-perfsonar-mesh is an excellent example 
> of mixing multiple meshconfigs and manually defined dashboards.We kept it 
> public so people could see it as an example for cases like this, so glad 
> you found it. Feel free to use the config examples or scripts from that 
> repo to help, but they are certainly not required to do what you want. 
> 
> Hope that helps,
> Andy
> 
> 
> 
> 
> 
> 
> On Tue, Dec 8, 2015 at 2:04 PM, Christopher J. Tengi 
> <>
>  wrote:
> 
> I am looking to add some grids and/or dashboards to our MaDDash host. Right 
> now, we have MaDDash running on our MA, with the display being generated by 
> a mesh.json file. We would like to be able to do 2 things: add support for 
> another mesh (or 2), and add some temporary grids for short-to-medium term 
> tests that are not part of a mesh. These tests are defined “by hand” on 
> individual perfSONAR nodes. 
> 
> I am assuming that the latter would be considered manual configuration ( 
> :-) ), and the http://software.es.net/maddash/mesh_config.html says that 
> "all existing groups, grids, checks, dashboards and general properties that 
> are defined outside the mesh will be preserved.” Section 2.3 of 
> https://www.perfsonar.net/deploy/installation-and-configuration/maddash-howto/
>  implies that if there is an existing maddash.yaml before the script is run 
> to process the mesh file, that the configuration in the original 
> maddash.yaml will be preserved. If that is the case, can I just try editing 
> the existing maddash.yaml file and have my changes preserved the next time 
> the mesh.json changes and the cron job is run to update maddash.yaml? 
> 
> For the latter case of creating multiple dashboards from multiple mesh.json 
> files, it looks like I just need to add more “<mesh>” blocks to 
> /opt/perfsonar_ps/mesh_config/etc/gui_agent_configuration.conf. Would it 
> make sense to create a “fake” mesh for my short-to-medium term tests, 
> specifying that there is no mesh agent on any of the end-points? 
> 
> Hmmm… Continuing my Google searches while composing this message, I just 
> found https://libraries.io/github/esnet/esnet-perfsonar-mesh, which says 
> that manual configuration can be placed in maddash.yaml.template. Is there 
> similar functionality outside of ESnet’s perfSONAR/MaDDash deployment? 
> 
> 
> Thanks, 
> /Chris 
> 
> 
>