Skip to Content.
Sympa Menu

perfsonar-user - Re: [perfsonar-user] Scheduled test/One way latency

Subject: perfSONAR User Q&A and Other Discussion

List archive

Re: [perfsonar-user] Scheduled test/One way latency


Chronological Thread 
  • From: David Koski <>
  • To: Andrew Lake <>
  • Cc: "" <>
  • Subject: Re: [perfsonar-user] Scheduled test/One way latency
  • Date: Wed, 30 Jul 2014 09:48:41 -0400

Yes thank you.  That's exactly what I needed to know.

       Dave


------------

David Koski

GCFA, GISP, GSEC, RHCE

ITS ISO Communications Systems and Data Centers Operations

734-647-8993 (Office)                                                                                     734-678-5824 (Mobile)
Description: cid:3337913627_541106



On Wed, Jul 30, 2014 at 9:36 AM, Andrew Lake <> wrote:
Hi David,

The OWAMP tests setup through the GUI create a local test to itself by default to the valid addresses on your host. The first two are just those tests which is why both the source and destination are your local address. It has to do IPv4 and IPv6 separately which is why there are two. I believe the idea was that they provide a baseline for measurements to external hosts (i.e. if you are seeing loss or some other strangeness in a local test, there's a good chance you have some type host problem). Does  that clarify things?

Thanks,
Andy


On Jul 30, 2014, at 8:56 AM, David Koski <> wrote:

I know I've already asked this question but thought I'd rephrase it to see if someone has any input.

I have a fresh install of perfsonar-ps loaded on a test system.  There have been no configuration changes to the system after running the netboot and the system only has 1 interface active.

When I add a One Way Latency test with a single test member (Using Scheduled Tests in the PerfSONER web page of the test host), what I end up getting is three graphs under active tests.

Source DestinationBidirectionalForward Direction Loss (Past 30 minutes)Reverse Direction Loss (Past 30 minutes)Graph
141.213.135.73 (141.213.135.73)141.213.135.73 (141.213.135.73)No 0.00%*
2607:f018:699:fffa:2e44:fdff:fe87:578c (2607:f018:699:fffa:2e44:fdff:fe87:578c) 2607:f018:699:fffa:2e44:fdff:fe87:578c (2607:f018:699:fffa:2e44:fdff:fe87:578c)No0.00%*
ntap-core.umnet.umich.edu (198.108.3.33)
141.213.135.73 (141.213.135.73) No0.00%*


The external_address for the test system is IPv4 141.213.135.73 and IPv6 2607:f018:699:fffa:2e44:fdff:fe87:578c.  So I'm getting 2 graphs testing internal to the perfsonar-ps for eth0 on the test system.


The last graph is what I expected which is to the test member I added.


The question I had is this normal behavior to see three graphs show up.  One for the test member, and two essentially looping back internally on the PerfSONAR server itself.


It would seem to me this would be expected behavior since the system I'm running is clean right out of the box, but I wanted to verify that in fact the PerfSONAR is using it's external address interface as a test member as well.


If this isn't normal behavior out of the box, does anyone know what might be wrong?


Just trying to put some explanation together for management and don't have much exposure to the perfsonar load yet.

   

     Thanks


                 David






------------

David Koski

GCFA, GISP, GSEC, RHCE

ITS ISO Communications Systems and Data Centers Operations

734-647-8993 (Office)                                                                                     734-678-5824 (Mobile)
<image001.png>









Archive powered by MHonArc 2.6.16.

Top of Page