perfSONAR User Q&A and Other Discussion

[Dan Doyle <>]

The throughput set of tools, of which nuttcp is a part of in pscheduler, store their entire stdout/stderr in the run result in the "diags" section. This can be pulled out via the pscheduler API for any completed run.

Here's an example based on a dev host that I have. Off the command line you might do something like:

`pscheduler task --tool nuttcp throughput <whatever arguments you want>`

It will emit something like...

Next scheduled run:

https://perfsonar-dev.grnoc.iu.edu/pscheduler/tasks/a735f4af-f8f1-4408-bd20-1af6d2d9b822/runs/803bd1e8-f0f2-454d-9161-0fb5ecc63e84

Starts 2018-01-30T21:12:10Z (~6 seconds)

Ends   2018-01-30T21:12:26Z (~15 seconds)

That URL is what you want. Once the run has finished, you can pull down the results of it with the tool of your choice and you will get the JSON result for it. You probably want to look for the "diags" key under "result-merged", which is the raw output of the underlying command.

A somewhat raw way of looking at it in your terminal would be:

`curl -k https://perfsonar-dev.grnoc.iu.edu/pscheduler/tasks/a735f4af-f8f1-4408-bd20-1af6d2d9b822/runs/803bd1e8-f0f2-454d-9161-0fb5ecc63e84| jq .`

Dan Doyle
GlobalNOC Software Developer
1-812-856-3892

On Jan 30, 2018, at 4:01 PM, Alan Whinery <> wrote:

It seems like I've heard about something like this during a Mark Feit
talk --

Can I run (for example)

    nuttcp -v -i1 <host>

 with pscheduler and then access the stdout from nuttcp after?

Attachment: smime.p7s
Description: S/MIME cryptographic signature