perfSONAR User Q&A and Other Discussion

[Zhi-Wei Lu <>]

Hi Andrew,

 

Here are the result of my ipv6 owamp ping to and from one of the hosts, either as sender or receiver, does it mean owampd is listening on my host?

 

bwping -T owamp -c 2607:f810:330:1ffe::f -s 2607:4000:500::5

bwping: Using tool: owamp

 

SENDER START

Approximately 13.3 seconds until results available

 

--- owping statistics from [perf-scidmz-data.cac.washington.edu]:5736 to [2607:f810:330:1ffe::f]:5888 ---

SID:       0000000fda9be4c0bf27af24653c76a5

first:     2016-03-22T08:35:31.125

last:      2016-03-22T08:35:38.676

10 sent, 0 lost (0.000%), 0 duplicates

one-way delay min/median/max = 10.3/10.4/10.5 ms, (err=0.732 ms)

one-way jitter = 0.2 ms (P95-P50)

Hops = 6 (consistently)

no reordering

 

 

SENDER END

 

bwping -T owamp -s 2607:f810:330:1ffe::f -c 2607:4000:500::5

bwping: Using tool: owamp

 

SENDER START

Approximately 13.3 seconds until results available

 

--- owping statistics from [melange-v6.noc.ucdavis.edu]:5892 to [2607:4000:500::5]:5741 ---

SID:       00000005da9be4ff602a07f3c5acf3e8

first:     2016-03-22T08:36:32.874

last:      2016-03-22T08:36:45.468

10 sent, 0 lost (0.000%), 0 duplicates

one-way delay min/median/max = 16.7/16.9/17.1 ms, (err=0.732 ms)

one-way jitter = 0.2 ms (P95-P50)

Hops = 6 (consistently)

no reordering

 

 

SENDER END

 

Zhi-Wei Lu

IET-CR-Network Operations Center

University of California, Davis

(530) 752-0155

 

From: Andrew Lake [mailto:]
Sent: Tuesday, March 22, 2016 6:35 AM
To: Zhi-Wei Lu <>;
Subject: Re: [perfsonar-user] Issue with perfsonar ipv6 and two NIC adapters

 

Hi,

 

Is owampd running on 2607:f810:330:1ffe::f and is it listening on the v6 interface? Either of those not holding true would explain what you are seeing. If you run "bwping -T owping” it will spawn its own owampd, which would explain why it works in that case but not for your regular tests which use the powstream client and assume owampd is already running on the other side.

 

Thanks,

Andy

 

 

On March 21, 2016 at 6:08:28 PM, Zhi-Wei Lu () wrote:

Hi all,

 

Using Linux policy routing, I was able to configure two adapters to perform bandwidth and owamp tests separately, since owamp is rather sensitive to network jitter and mixing bandwidth test and owamp is not a great idea.

 

The CENIC maddash dashboard shows that our host mélange.noc.ucdavis.edu performs fine with two NICs on IPV4.

https://ps-dashboard.cenic.net/maddash-webui/index.cgi?dashboard=CENIC_10G-connected

 

A while ago, I setup our IPV6 for the 10G bandwidth test with rest of the CENIC grid, it worked fine until I setup an IPV6 address for the owamp test on the second adapter, it interfered with our main IPV6 bandwidth test, thus I turned off IPV6 address on the second adapter.  With the impetus of CENIC conference here at UCD, I started to configure policy routing for our IPV6 on these two adapters.  I verified the IPV6  ping6, traceroute6, and bwping –T owamp to a few CENIC hosts. 

 

It took a long while for the maddash graph to turn HALF green for our host (mélange.noc.ucdavis.edu, 2607:f810:330:1ffe::f) to other CENIC grid hosts.  I wonder why I don’t get the FULL green on the maddash graph, since I was able to bwping to CENIC hosts both ways.  Does the other sites disable IPV6 owping tests to us since we were off the radar screen for a long time?

 

Thank you.

 

Zhi-Wei Lu

IET-CR-Network Operations Center

University of California, Davis

(530) 752-0155