perfSONAR User Q&A and Other Discussion

[Andrew Lake <>]

Hi Murilo,

What happens when you run “owping -S 200.237.196.43 200.237.196.45” from the command line of the CentOS host? Looks like pscheduler is correctly starting the powstream process but once its launched the OWAMP test is not working. The way powstream works is it get launched and then spawns two child processes that actually run the tests (hence the 3 processes for the same test on your working Debian host). The fact that CentOS is only showing one confirms its getting launched but for some reason the child processes are not getting created. Lots of time its just firewall issues, but the owping should give you a better idea hopefully. You can also try looking in /var/log/pscheduler/pscheduler.log for any references to powstream in error messages.

Thanks,

Andy

On August 11, 2017 at 10:13:23 AM, Murilo Vetter () wrote:

Hi list,

I am running a lab here in RNP, using 4 perfsonar nodes, 2 with CentOS and 2 with Debian.

I scheduled 2 meshes, one with CentOS hosts and another with Debian hosts, running powstream tests and iperf3 tcp tests.

All is working, except the powstream tests between CentOS. I couldn't figure out what could be.

When I check powstream process on CentOS, it appears like this:

# ps aux | grep pow
pschedu+  6756  0.0  0.1 266752 18940 ?        S    11:00   0:00 /usr/bin/python /usr/libexec/pscheduler/commands/../classes/tool/powstream/run
pschedu+  8248  0.0  0.0   7564   936 ?        S    11:03   0:00 /usr/bin/powstream -p -d /var/lib/pscheduler/tool/powstream/18bf28d6-fdb1-431a-8262-fc4e3614e321-2017Aug11T140001195158 -s 0 -c 600 -i 0.1 -t -S 200.237.196.43 [200.237.196.45]:861
root      8254  0.0  0.0 112668   964 pts/0    S+   11:03   0:00 grep --color=auto pow

When I check powstream process on Debian, it appears like this:

# ps aux | grep pow
pschedu+  5596  0.0  0.2 128864 42712 ?        S    Aug10   0:20 /usr/bin/python /usr/lib/pscheduler/commands/../classes/tool/powstream/run
pschedu+  5606  0.0  0.0  13852  2548 ?        S    Aug10   0:04 /usr/bin/powstream -p -d /var/lib/pscheduler/tool/powstream/66b434fa-041f-49da-9247-7a75ffd10d1a-2017Aug10T201539801126 -s 0 -c 600 -i 0.1 -t -S 200.237.196.42 [200.237.196.44]:861
pschedu+ 15430  0.0  0.0  13848   336 ?        Ss   11:07   0:00 /usr/bin/powstream -p -d /var/lib/pscheduler/tool/powstream/66b434fa-041f-49da-9247-7a75ffd10d1a-2017Aug10T201539801126 -s 0 -c 600 -i 0.1 -t -S 200.237.196.42 [200.237.196.44]:861
pschedu+ 15521  0.0  0.0  13852   340 ?        Ss   11:08   0:00 /usr/bin/powstream -p -d /var/lib/pscheduler/tool/powstream/66b434fa-041f-49da-9247-7a75ffd10d1a-2017Aug10T201539801126 -s 0 -c 600 -i 0.1 -t -S 200.237.196.42 [200.237.196.44]:861
root     15556  0.0  0.0  12728  2220 pts/0    S+   11:09   0:00 grep pow

I install in this hosts the "perfsonar-testpoint" packages.

In CentOS when I run pscheduler monitor it shows like this:

2017-08-11T11:00:01-03:00 Running      latencybg --dest 200.237.196.45 --packet-padding 0 --source 200.237.196.43 --packet-interval +

On Debian hosts it is running as expected.

Do you have any clue about this problem?

Regards,

--

Murilo Vetter
Analista REMEP-FLN/PoP-SC
PoP-SC - Ponto de Presença da RNP em Santa Catarina
REMEP-FLN - Rede Metropolitana de Educação e Pesquisa da Região de Florianópolis
RNP - Rede Nacional de Ensino e Pesquisa
UFSC - Universidade Federal de Santa Catarina
SETIC - Superintendência de Governança Eletrônica e Tecnologia da Informação e Comunicação
Fundações de Apoio FEESC e FEPESE
--/--

www.pop-sc.rnp.br
remep.pop-sc.rnp.br
CHAMADOS:
+55(48)3721-3000
INOC: 11242*100