perfSONAR User Q&A and Other Discussion

[Ignacio Peluaga Lozada <>]

Hello Ivan,

thanks for your message. I was able eventually to achieve that but not via 
toolkit UI, instead I used pSConfig. Basically, in the pSConfig file, I 
defined inside the 'tests' section for the throughput test: dest, source-node 
and duration. Therefore the other "conflicting" options are not present in 
the test. 

Once I publish and add that json file on the server, I do 'pscheduler 
monitor' and see the test runs successfully. I did the same thing for host B, 
with its private IP as source-node, all works fine. In conclusion, I can 
schedule these throughput tests:

- Host A -> Host B (and see results in hosts A's toolkit UI)
- Host B -> Host A (and see results in hosts B's toolkit UI) 

This is good already, but it would be very convenient if I could see the 
results of both directions on a single graph (Forward and Reverse) or at 
least two different graphs but in the same toolkit UI.

Therefore, I created a single pSConfig file and defined there two tests and 
tasks: Host B -> Host A and the same one using "reverse": true. I specified 
for both tasks' the same MA endpoint in archives: the one in host B. After 
publishing and adding that file on host B, I do 'pscheduler monitor' on both 
nodes and see both tests run correctly.

However, the graph on host B's toolkit UI only shows Forward results, not 
Reverse. Do you have any idea what could be the reason why the reverse 
measurements are not being sent to host B's MA?

Thanks in advance!

Regards,
Ignacio

________________________________________
From:  
[] on behalf of Garnizov, Ivan 
[]
Sent: 27 January 2021 16:13
To: Ignacio Peluaga Lozada; 
Subject: RE: [perfsonar-user] Throughput test via Toolkit

Hello Ignacio,

The short answer is it is not possible...
It is also not possible, nor advisable to run underspecified tests.
All in all even if the Toolkit high level service does not set them, then it 
would be the lower level setting them automatically for you.

Instead I would strongly recommend to see in the task details of the 
successful run, what exactly the parameter specifications was and then try to 
reproduce it.
IMO It would be much easier if you make several tests on the CLI to find / 
match the correct set of parameters for a complete spec of the test, which 
succeeds just the way you managed.

It is interesting also you managed to have a successful test with a system 
behind NAT, which means the scheduler on the B host has recognized the IP 
address you are calling for a test as own address.


Regards,
Ivan Garnizov

GEANT WP6T3: pS development team
GEANT WP7T1: pS deployments GN Operations
GEANT WP9T2: Software governance in GEANT



-----Original Message-----
From:  
[] On Behalf Of Ignacio Peluaga 
Lozada
Sent: Wednesday, January 27, 2021 12:47 PM
To: 
Subject: [perfsonar-user] Throughput test via Toolkit

Hello all,

I have a question regarding scheduling throughput measurements between two 
hosts, A and B, using the toolkit. Host B is behind NAT, host A is not.

Via host A's toolkit I have scheduled a throughput test to host B, but no 
results are reported.

Therefore I ssh'd into host A and run 'pscheduler monitor' and saw there 
indeed the throughput task as 'Failed'. I run the command there shown and it 
fails (I get 'Run did not complete: Failed'), however if I remove all the 
options except '--dest' it works.

So my question is, is there anyway to avoid having the toolkit set those 
options (source, source-node, dest-node, etc)?

Thanks.

Regards,
Ignacio