perfSONAR User Q&A and Other Discussion

[Andrew Lake <>]

Hi Casey,

Looks like the problem is that the test is getting stored in the same archive twice. If you look in the "archives" section of https://ps-ku-bw.perfsonar.kanren.net/pscheduler/tasks/bd813fba-cf74-4bff-8a5c-3458f9652931?pretty you will see 3 entries: one goes to a local archive on ps-ku-bw (which is the archive you use in the correct graph) and two to the central  one on ps-dashboard (which is the on in use for the double graph). Normally psconfig removes duplicate entries, but the URL of one has a trailing slash and the other doesn't, so it leaves them separate since there is a difference in the definitions. 

Likely the cause is that you have "configure-archives" set and a default archive pointing at that host each with a different URL variation. You can check if configure-archives is set by looking at the out put of the "psconfig remote list" command. Likely the quickest fix is to delete the archive file under /etc/perfsonar/psconfig/archives.d that has the duplicate definition as your central psConfig template file. My guess is the file will be called  something like  /etc/perfsonar/psconfig/archives.d/ps-dashboard.kanren.net.json if the migration script put it there. You will likely want to leave the local archive there, so don't delete  /etc/perfsonar/psconfig/archives.d/esmond_local.json since I am guessing that is not the problem file. That approach will leave the definition in your central file but you could also do the reverse if needed and disable configure-archives and keep the include file. Also I am assuming some things here so let me know if you can't find the file or configure-archives is not set.

Thanks,
Andy

On Fri, Aug 31, 2018 at 10:47 AM, Casey Russell <> wrote:

Group,

 BEHOLD!  I have invented the NETWORK DOUBLER!  

Ok, thank you for indulging my attempt at humor.  I've fixed one problem and I'm moving on to another that is vexxing me since last week ( I think it happened when I converted my older mesh config and maddash.yaml files using the psconfig agents.).  Maddash now reports twice the bandwidth that the toolkit host does for the same test.  See here:

http://ps-dashboard.kanren.net/maddash-webui/details.cgi?uri=/maddash/grids/KanREN-PS+-+IPv4+iPerf+Bandwidth+-+UDP+-+Throughput/ps-ku-bw.perfsonar.kanren.net/ps-esu-bw.perfsonar.kanren.net/Throughput

vs:

http://ps-ku-bw.perfsonar.kanren.net/perfsonar-graphs/?source=164.113.32.145&dest=164.113.32.109&url="http%3A%2F%2Fps-ku-bw.perfsonar.kanren.net%2Fesmond%2Fperfsonar%2Farchive%2F#start=1535120931&end=1535725731&summaryWindow=3600&timeframe=1w

In fact, Maddash is reporting 2x the bandwidth that the NIC on the testing machines are capable of.

Because it's only affecting Maddash, I suspected the problem is with the Maddash.yaml file, or something else related to the dashboard.  But if there's a problem with the maddash.yaml, I can't see it from reading it.  Anyone have any ideas where I should be looking,  or tips as to what might cause this?  I can supply maddash.yaml or other config files offline if it'll help.

Sincerely,

Casey Russell

Network Engineer

785-856-9809

2029 Becker Drive, Suite 282
Lawrence, Kansas 66047

--
To unsubscribe from this list: https://lists.internet2.edu/sympa/signoff/perfsonar-user