perfSONAR User Q&A and Other Discussion

[Peter van Heusden <>]

Ok, I found the cause of the problem!

Somehow the web user interface is setting the TOS Bits field of the bandwidth test I had set up to "NaN". This then leads to the following lines (around line 1243) of bwmaster.pl triggering and adding a "-S NaN" to the bwctl command line. Then bwctl of course fails and is restarted... etc etc.

    push @cmd, ( "-S", $val ) if (
        $val = $conf->get_val(
            TESTSPEC => $ms->{'TESTSPEC'},
            ATTR     => 'BWTosBits'
        )
    );

I manually removed the BWTosBits entry in /opt/perfsonar_ps/perfsonarbuoy_ma/etc/owmesh.conf and the restart loop has now stopped.

Thanks,
Peter

On 23/05/2013 22:48, Peter van Heusden wrote:

Yes, those work fine. E.g.:

[root@ps sysconfig]# bwctl -c 192.168.2.132 -s 192.168.2.104
bwctl: Using tool: iperf
bwctl: 15 seconds until test results available

RECEIVER START
bwctl: exec_line: iperf -B ps2.sanbi.ac.za -s -f b -m -p 5176 -t 10
bwctl: start_tool: 3578330236.026986
------------------------------------------------------------
Server listening on TCP port 5176
Binding to local address ps2.sanbi.ac.za
TCP window size: 87380 Byte (default)
------------------------------------------------------------
[ 15] local 192.168.2.132 port 5176 connected with 192.168.2.104 port 5176
[ ID] Interval       Transfer     Bandwidth
[ 15]  0.0-10.0 sec  530448384 Bytes  422283774 bits/sec
[ 15] MSS size 1448 bytes (MTU 1500 bytes, ethernet)
bwctl: stop_exec: 3578330249.065921

RECEIVER END

Is there any other log file that I could be looking in? The messages in /var/log/messages don't explain *why* bwctl is being restarted, also there is a bwctl process running the entire time, so I'm not sure *what* is being restarted!

Thanks!
Peter
On 23/05/2013 22:28, Aaron Brown wrote:

Hey Peter,

Can you run bwctl tests by hand?

Cheers,

Aaron

On May 23, 2013, at 4:27 PM, Peter van Heusden <> wrote:

I made that change, and added tock.meraka.csir.co.za - a stratum 1 server that is about 1000 miles away. ntpq -p now shows:

     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*zibbi.meraka.cs 238.72.153.243   2 u   40   64    1   72.996   -3.658   6.944
+firewall.sanbi. 41.73.38.11      3 u   22   64    7    0.070   -4.745   0.173
-ntp0.za.uu.net  216.171.120.36   3 u   20   64    7    5.677    0.208  15.795
-ntp2.is.co.za   146.64.58.41     2 u   21   64    7    5.525   -9.424  16.250
+tock.meraka.csi .PPS.            1 u    -   64   15   73.582   -3.607  21.681


and bwmaster.pl is still restarting.

:(

On 23/05/2013 21:46, Pedro Queirós wrote:

Try this:

server ntp1.meraka.csir.co.za iburst minpoll 4 maxpoll 6

server ntp.sanbi.ac.za iburst minpoll 4 maxpoll 6
server ntp0.za.uu.net iburst minpoll 4 maxpoll 6
server ntp2.is.co.za iburst minpoll 4 maxpoll 6

Asides from that, it looks good. If possible, try to have access 

to good (e.g. low delay) NTP stratum 1 server near your network.

If the bwmaster.pl continues restarting, I'd suggest looking into

something else - let us know about!

Ah, don't forget to restart ntpd after changing the config file!

Pedro

On Thu, May 23, 2013 at 8:31 PM, Peter van Heusden <> wrote:

logfile /var/log/ntpd
driftfile /var/lib/ntp/ntp.drift
statsdir  /var/lib/ntp/
statistics loopstats peerstats clockstats
filegen loopstats file loopstats type day enable
filegen peerstats file peerstats type day enable
filegen clockstats file clockstats type day enable

# You should have at least 4 NTP servers

server ntp1.meraka.csir.co.za iburst
server ntp.sanbi.ac.za iburst
server ntp0.za.uu.net iburst
server chronos.es.net iburst
server ntp2.is.co.za iburst

thanks!

On 23/05/2013 21:17, Pedro Queirós wrote:

Peter, from the ntpq -p output I can see your NTP config is faulty.

Can you provide the /etc/ntp.conf file?

Kind Regards,

Pedro

On Thu, May 23, 2013 at 8:04 PM, Jason Zurawski <> wrote:

While not true in this case that you sent - static lists fall into disrepair frequently.  A couple of months back we tried this in the APAN region and found about 1/5 still worked well after being posted to an 'official' wiki.

It doesn't hurt to ask those on the front lines that are adopting pS for some insider info - that request applies to all, even if they ate not in South Africa.

Thanks;

-jason

On May 23, 2013, at 11:58 AM, Michael Sinatra <> wrote:

> On 05/23/2013 11:44, Jason Zurawski wrote:
>> Hey Peter;
>>
>> I would nuke the ESnet server too - NTP works best when the servers are within the same timezone/continent, it gives the algorithms a level playing field to choose from.
>
> It's not really the case that a server outside of the timezone or continent matters.  The key is whether it's more likely than not that there will be asymmetry in the path to/from a given NTP server.  I don't really see much evidence of that with any of the servers in the list, except for possibly ntp.mtnbusiness....
>
>
>> A related note is that we are always looking to add new servers into our list around the world - if you know of 'open' clocks in the region that we can front-load into the list, that can be arranged.
>
> Well, there's a published list of them here:
>
> http://support.ntp.org/bin/view/Servers/WebHome
>
> :)
>
> michael
>

ESnet/Internet2 Focused Technical Workshop
Network Issues for Life Sciences Research
July 17 - 18, 2013, Berkeley CA
http://events.internet2.edu/2013/ftw-life-sciences/