perfSONAR User Q&A and Other Discussion

[Peter Jacobs <>]

I wanted to report that the issue of one-way latency tests failing between 
two LAT nodes was caused by MTU mis-match along the path of network 
equipment.  Once this was sorted out, the bidirectional tests started 
working.  That was the last thing I was expecting to cause an application 
level problem, but we are working now

Thanks for the suggestions below.

Regards


Peter Jacobs
University of New Brunswick
Canada



-----Original Message-----
From: Jason Zurawski 
[mailto:]
 
Sent: April-03-14 1:05 PM
To: Peter Jacobs
Cc: 

Subject: Re: [perfsonar-user] One Way latency test failure

Hi Peter;

One thing to try would be to disable iptables and ip6tables between the hosts 
if you have not done so - that error typically means one of two things:

1) Firewalls (of some sort) blocking a control or data connection.  Control 
would be port 861 (TCP) and data is defined by the 'peer_ports' in 
/etc/owampd/owamd.conf (it will be UDP traffic).  

2) MTU issues could also cause this

3) Ensure NTP is functional, and the error is small:

> [zurawski@newy-owamp
>  ~]$ /usr/sbin/ntptime 
> ntp_gettime() returns code 0 (OK)
>   time d6e804f5.9530f02c  Thu, Apr  3 2014  9:04:05.582, (.582778516),
>   maximum error 2256 us, estimated error 3 us, TAI offset 0
> ntp_adjtime() returns code 0 (OK)
>   modes 0x0 (),
>   offset -7.364 us, frequency -53.877 ppm, interval 1 s,
>   maximum error 2256 us, estimated error 3 us,
>   status 0x2001 (PLL,NANO),
>   time constant 5, precision 0.001 us, tolerance 500 ppm,

Error should be < 1ms if we can get it to that level, although OWAMP should 
still continue to function when error approaches the threshold of seconds.  

Beyond that we don't have many other suggestions unfortunately, 

Thanks;

-jason

> From: Peter Jacobs 
> <>
> Subject: [perfsonar-user] One Way latency test failure
> Date: April 1, 2014 at 3:54:17 PM EDT
> To: 
> ""
>  
> <>
> 
>  
> I am having issues with my one-way latency tests on node 
> lat-gp-ftn.nbren.ca to one destination, but other destinations are fine.  
> Based on the information and error below, where should I start looking to 
> solve the problem? 
>  
> My node, lat-gp-ftn.nbren.ca is running a fresh net install of 3.3.2  and 
> the other, lat-hlfx1pfs1.canarie.ca, is running 3.3.1  They can see each 
> other and there are no firewalls between the nodes.   All other scheduled 
> latency tests work fine on both nodes.
>  
>  
> On this screen there is no reverse direction stat on the 5th line in the 
> table.  http://lat-gp-ftn.nbren.ca/serviceTest/index.cgi?eventType=owamp
>  
> but on the other node looking back (line 6) is okay.  
> http://lat-hlfx1pfs1.canarie.ca/serviceTest/index.cgi?eventType=owamp
>  
> Here are the results of a manual owping command...
>  
> First from lat-gp-ftn.nbren.ca which partially fails to the destination
>  
> [lat-gp-ftn ~]# owping lat-hlfx1pfs1.canarie.ca
>  
> Approximately 13.1 seconds until results available
>  
> owping: FILE=protocol.c, LINE=1270, _OWPReadTestRequest: Unable to read 
> from socket.
> owping: FILE=owping.c, LINE=1837, Unable to fetch data for 
> sid(cdbd209ad6e3de34d4c21ec042ff4e96)
>  
> --- owping statistics from [lat-hlfx1pfs1.canarie.ca]:8785 to 
> [lat-gp-ftn.nbren.ca]:8955 ---
> SID:    c6a41dcad6e3de34dd29e8d81aaece90
> first:  2014-03-31T09:29:41.973
> last:   2014-03-31T09:29:51.357
> 100 sent, 0 lost (0.000%), 0 duplicates
> one-way delay min/median/max = 4.64/4.8/4.77 ms, (err=0.402 ms)
> one-way jitter = 0 ms (P95-P50)
> Hops = 2 (consistently)
> no reordering
>  
>  
> Then from the other node back to my node:
>  
> [lat-hlfx1pfs1 etc]#  owping lat-gp-ftn.nbren.ca
>  
> Approximately 13.2 seconds until results available
>  
> --- owping statistics from [lat-hlfx1pfs1.canarie.ca]:8909 to 
> [lat-gp-ftn.nbren.ca]:8859 ---
> SID: c6a41dcad6e578ab32bfa21fb729648b
> first: 2014-04-01T14:41:00.382
> last: 2014-04-01T14:41:11.102
> 100 sent, 0 lost (0.000%), 0 duplicates
> one-way delay min/median/max = 4.16/4.3/4.27 ms, (err=0.162 ms)
> one-way jitter = 0 ms (P95-P50)
> Hops = 2 (consistently)
> no reordering
>  
> --- owping statistics from [lat-gp-ftn.nbren.ca]:8960 to 
> [lat-hlfx1pfs1.canarie.ca]:8917 ---
> SID: cdbd209ad6e578ab340d3b7c2d302acd
> first: 2014-04-01T14:41:00.439
> last: 2014-04-01T14:41:08.994
> 100 sent, 0 lost (0.000%), 0 duplicates
> one-way delay min/median/max = 3.98/4.1/4.09 ms, (err=0.162 ms)
> one-way jitter = 0 ms (P95-P50)
> Hops = 2 (consistently)
> no reordering
>  
>  
> Thanks for any suggestions.
>  
>  
> Peter E. 
> 
> UNB - Information Technology Services