perfSONAR User Q&A and Other Discussion

[John Simpkins <>]

To update everyone on this thread, I installed the PS 4.0 RC1 toolkit installed directly from the ISO. I believe the earlier messages I posted were due to not having the appropriate bwctl-server.limits rules applied. Currently, it appears that tests are working in one direction but not the other. Here are the messages being observed:

Oct 17 08:56:48 perfsonar4-lab-ma journal: scheduler DEBUG    Posting https://perfsonar4-lab-ma/pscheduler/tasks/c7b8e0fd-8363-4a19-b7ba-95db78aadc3a at 2016-10-17 08:56:58.289267-04:00

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Program threw an exception after 0:00:00.022652

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Program threw an exception after 0:00:00.022652

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Exception: URLException: https://perfsonar4-lab-ma/pscheduler/tasks/c7b8e0fd-8363-4a19-b7ba-95db78aadc3a: 400: Error: ('Connection aborted.', gaierror(-2, 'Name or service not known'))#012Traceback (most recent call last):#012  File "/usr/lib/python2.7/site-packages/pscheduler/saferun.py", line 41, in safe_run#012    function()#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 430, in <lambda>#012    pscheduler.safe_run(lambda: main_program())#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 394, in main_program#012    run_post(url, trynext, log)#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 96, in run_post#012    status, task = pscheduler.url_get(task_url_text, params={'detail': 1})#012  File "/usr/lib/python2.7/site-packages/pscheduler/psurl.py", line 57, in url_get#012    + ": " + text)#012URLException: https://perfsonar4-lab-ma/pscheduler/tasks/c7b8e0fd-8363-4a19-b7ba-95db78aadc3a: 400: Error: ('Connection aborted.', gaierror(-2, 'Name or service not known'))

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Exception: URLException: https://perfsonar4-lab-ma/pscheduler/tasks/c7b8e0fd-8363-4a19-b7ba-95db78aadc3a: 400: Error: ('Connection aborted.', gaierror(-2, 'Name or service not known'))#012Traceback (most recent call last):#012  File "/usr/lib/python2.7/site-packages/pscheduler/saferun.py", line 41, in safe_run#012    function()#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 430, in <lambda>#012    pscheduler.safe_run(lambda: main_program())#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 394, in main_program#012    run_post(url, trynext, log)#012  File "/usr/libexec/pscheduler/daemons/scheduler", line 96, in run_post#012    status, task = pscheduler.url_get(task_url_text, params={'detail': 1})#012  File "/usr/lib/python2.7/site-packages/pscheduler/psurl.py", line 57, in url_get#012    + ": " + text)#012URLException: https://perfsonar4-lab-ma/pscheduler/tasks/c7b8e0fd-8363-4a19-b7ba-95db78aadc3a: 400: Error: ('Connection aborted.', gaierror(-2, 'Name or service not known'))

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Waiting 348.25 seconds before restarting

Oct 17 08:56:48 perfsonar4-lab-ma journal: safe_run/scheduler ERROR    Waiting 348.25 seconds before restarting

On Fri, Oct 14, 2016 at 2:39 PM John Simpkins <> wrote:

I paved over and have a working PS 4.0 RC1 toolkit now. Joined it to a 3.5 mesh, awaiting population of test results. If I understand correctly, pscheduler takes care of updating tests so they fire correctly. Does anyone know if setting "configure_archives 1" works in 4.0 so the toolkit will report to a MA running on a 3.5 host?

On Wed, Oct 12, 2016 at 9:29 AM Bill Owens <> wrote:

On Wed, Oct 12, 2016 at 01:19:16PM +0000, Mark Feit wrote:
> Bill Owens writes:
> >
> > It's whatever is currently in the 4.0RC repo. The command itself doesn't seem to have a version flag so I'm not sure how to get that.
> >
> >BTW, I did run 'pscheduler diags' and it filled a file with error logs, as well as throwing this:
> >
> >[owens@bss532 ~]$ sudo /usr/bin/pscheduler diags >pscheduler.diags.20161012.txt
> >Server returned status 500: column "task_cli" does not exist
> >LINE 10:                 task_cli
> >                        ^
> >
> >It's really not happy about task_cli! ;)
>
> That sounds to me like you’re running the unstable daily build or a hybrid of that and what was in RC1.  None of that was added until after RC1 was released.
>
> --Mark
>
>

I followed the instructions to install 4.0RC1 because I was having performance issues on my hardware (installing pS on my boxes caused an enormous drop in their bandwidth testing performance) and I was hoping that CentOS 7 would help. All I've done since then is run "yum update" a couple of times, in the hopes that whatever was broken and preventing my machine from running tests would be fixed by one of the updates. Here's what shows up on my box for perfsonar-toolkit:

Installed Packages
Name        : perfsonar-toolkit
Arch        : noarch
Version     : 4.0
Release     : 0.7.rc1.el7.centos
Size        : 6.2 M
Repo        : installed
From repo   : Internet2-staging
Summary     : perfSONAR Toolkit
URL         : http://www.perfsonar.net/
License     : Distributable, see LICENSE
Description : The perfSONAR Toolkit web GUI and associated services.

Is that the RC package, or something else?

Should I pave over this install and try the ISO instead?

Bill.