perfSONAR User Q&A and Other Discussion

["Victor Olifer" <>]

Hi,
When I run pscheduler with iperf2 a server always returns an error, like this:

[victor@ps-slough ~]$ pscheduler task --tool iperf2 throughput  --dest ps-
londhx1.ja.net
Submitting task...
Task URL:
https://localhost/pscheduler/tasks/0863553d-cd39-4746-a0e6-6d421a21270a
Running with tool 'iperf2'
Fetching first run...

Next scheduled run:
https://localhost/pscheduler/tasks/0863553d-cd39-4746-a0e6-6d421a21270a/runs/fbd373f4-fbde-4d23-8297-d46b56f20501
Starts 2019-10-09T13:55:43+01 (~8 seconds)
Ends   2019-10-09T13:55:58+01 (~14 seconds)
Waiting for result...

Run did not complete: Failed

Limit system diagnostics for this run:

  Hints:
    requester: ::1
    server: ::1
  Identified as everybody, local-interfaces
  Classified as default, friendlies
  Application: Hosts we trust to do everything
    Group 1: Limit 'always' passed
    Group 1: Want all, 1/1 passed, 0/1 failed: PASS
    Application PASSES
  Application: Defaults applied to non-friendly hosts
    Group 1: Limit 'innocuous-tests' failed: Passed but inverted
    Group 1: Limit 'throughput-default-tcp' passed
    Group 1: Limit 'throughput-default-udp' failed: UDP testing not allowed
    Group 1: Limit 'idleex-default' failed: Test is not 'idleex'
    Group 1: Want any, 1/4 passed, 3/4 failed: PASS
    Application PASSES
  Proposal meets limits
  Priority set to default of 0


Diagnostics from localhost:
  /usr/bin/iperf -p 5001 -c ps-londhx1.ja.net -t 10 -m

  ------------------------------------------------------------
  Client connecting to ps-londhx1.ja.net, TCP port 5001
  TCP window size: 14.1 MByte (default)
  ------------------------------------------------------------
  [  3] local 194.81.18.227 port 40422 connected with 194.83.97.209 port 5001
  [ ID] Interval       Transfer     Bandwidth
  [  3]  0.0-10.0 sec  11.5 GBytes  9.90 Gbits/sec
  [  3] MSS size 8140 bytes (MTU 8180 bytes, unknown interface)

Error from localhost:
  No error.
Diagnostics from ps-londhx1.ja.net:
  /usr/bin/iperf -s -p 5001
Error from ps-londhx1.ja.net:
  iperf2 returned an error: Process took too long to run.
No further runs scheduled.

The same test with iperf3 goes smoothly.
It is not the first time I experience this error, I and my colleagues from
JISC tested iperf2 under pscheduler many times and between different pairs of
hosts, and with pS 4.1 and pS 4.2.
Without pscheduler iperf2 runs OK:

[victor@ps-slough ~]$ iperf -c ps-londhx1.ja.net
------------------------------------------------------------
Client connecting to ps-londhx1.ja.net, TCP port 5001
TCP window size: 12.8 MByte (default)
------------------------------------------------------------
[  3] local 194.81.18.227 port 41200 connected with 194.83.97.209 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  9.30 GBytes  7.99 Gbits/sec
[victor@ps-slough ~]$

As under pscheduler iperf2 brings some plausible results it looks like at the
end of a session iperf2 returns something which pscheduler doesn't like.
Could it be fixed?

Thanks,
Victor