perfSONAR User Q&A and Other Discussion

[Andi Bergen <>]

Oh, ok. hmm.
Thanks for your help in diagnosing this, I really appreciate it.
Cheers,
Andi

On 3 September 2014 08:34, Aaron Brown 
<>
 wrote:
> Hey Andi,
>
> My guess is that the host itself is NAT’d from the outside world, but 
> there’s a 1:1 mapping between the external interface and the internal one. 
> Unfortunately, bwctl doesn’t currently support this model as it passes 
> around the IPs for the testing tools bind to.
>
> Cheers,
> Aaron
>
> On Sep 2, 2014, at 4:48 PM, Andi Bergen 
> <>
>  wrote:
>
>> Oh my, that is discouraging.
>> iperf -s -B 142.104.17.140 -p 6019
>> bind failed: Cannot assign requested address
>> (in fact: can't bind to any address).
>> Without the bind option it works and accepts connections from external
>> sources though.
>> Andi
>>
>>
>> On 2 September 2014 12:34, Aaron Brown 
>> <>
>>  wrote:
>>> Hi Andi,
>>>
>>> Can you run this on the server host?
>>>
>>> iperf -s -B 142.104.17.140 -p 6019
>>>
>>> Cheers,
>>> Aaron
>>>
>>> On Sep 2, 2014, at 2:34 PM, Andi Bergen 
>>> <>
>>>  wrote:
>>>
>>>> Hi Aaron,
>>>> 142.104.17.140 is the server host. Checking with netstat -nlp, port 6019 
>>>> is not used by anything.
>>>>
>>>> The public ip address of the VMs is visible and accessible from outside 
>>>> their private network.
>>>> Oh, I forgot to mention, sorry:
>>>> - if I run iperf on these ports, it works.
>>>> - Running iperf on the client and bwctl on the server does connect. 
>>>> Nothing happens, but then ctrl+c on client does show that a connection 
>>>> was made on the server (iperf).
>>>> - Running bwtcl on the client to iperf on the server works (the returned 
>>>> format is different).
>>>> - logging in from an external network to the public IP address works.
>>>>
>>>> We are using OpenStack to orchestrate the VMs.
>>>> It might be an issue with openstack, because a traceroute looks somewhat 
>>>> funny from 142.104.17.139 to the "server"
>>>> traceroute to 142.104.17.140 (142.104.17.140), 30 hops max, 60 byte 
>>>> packets
>>>> 1  142.104.17.140 (142.104.17.140)  95.876 ms  95.797 ms  95.732 ms
>>>> 2  * * *
>>>> 3  * * *
>>>> 4  * * *
>>>> 5  * * *
>>>> 6  * * *
>>>> 7  * * *
>>>> 8  * * *
>>>> 9  * * *
>>>> 10  * * *
>>>> 11  * * *
>>>> 12  * *^C
>>>>
>>>> I hope that helps,
>>>> Andi
>>>>
>>>>
>>>> 142.104.17.139
>>>>
>>>> On 2014-09-02 11:14 AM, "Aaron Brown" 
>>>> <>
>>>>  wrote:
>>>> Hi Andi,
>>>>
>>>> I see this error in the server logs:
>>>>
>>>>> Sep  2 13:01:05 centos bwctld[122131]: FILE=endpoint.c, LINE=319, 
>>>>> bind([142.104.17.140]:6019): Cannot assign requested address
>>>>
>>>>
>>>> Does the 142.104.17.140 host have the address 142.104.17.140 configured 
>>>> on it, or is it behind a NAT or something? I’m not sure why it wouldn’t 
>>>> be able to bind to that address.
>>>>
>>>> Cheers,
>>>> Aaron
>>>>
>>>> On Sep 2, 2014, at 1:40 PM, Andi Bergen 
>>>> <>
>>>>  wrote:
>>>>
>>>>> Hi,
>>>>> I am having problems getting the bwctl to run successfully between two
>>>>> of my install locations.
>>>>>
>>>>> I followed the steps outlined here:
>>>>> https://fasterdata.es.net/performance-testing/perfsonar/ps-howto/level-1/
>>>>> and have this installed on two VMs.
>>>>>
>>>>> I downloaded the limits file from:
>>>>> http://stats.es.net/sample_configs/bwctld.limits
>>>>> (which does include a /16 address for our institution).
>>>>>
>>>>> What doesn't work is running
>>>>> bwctl -vv -c 142.104.17.140:4823 -f m -x -a 100
>>>>> or bwctl -vv -c 142.104.17.140 -f m -x -a 100
>>>>>
>>>>> However, running this TO any other machine (I found listed here
>>>>> http://stats.es.net/ServicesDirectory/#search=es) works.
>>>>>
>>>>> It also works just fine when using the internal IP addresses of these
>>>>> machines since they are on the same network.
>>>>>
>>>>> I checked that bwctl is bound on the external interface. And
>>>>> connections are going through, but then fail (please see attached
>>>>> /var/log/messages files for both machines).
>>>>>
>>>>> All UDP and TCP ports are open for machines in the same security
>>>>> group, as well as all ports I could find in the docs are open
>>>>> otherwise. The ports mentioned in the log files are opened.
>>>>>
>>>>> Attached are the /var/log/messages from the VM that is running the
>>>>> test (client.log) and the machine that functions as the "server"
>>>>> (server.log).
>>>>>
>>>>> Did I follow the wrong instructions, or am I missing something
>>>>> obvious? I am somewhat at a loss for what to check next...
>>>>>
>>>>> Thank you,
>>>>> Andi
>>>>> ps.
>>>>> Command line output from "client" bwctl -vv -c 142.104.17.140:4823 -f
>>>>> m -x -a 100
>>>>> (at a different time than the attached log files, but it is always the
>>>>> same except the time stamps of course):
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=2931, Using 10.6.9.19 as the address
>>>>> for local sender
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=2931, Using 142.104.17.140:4823 as
>>>>> the address for remote receiver
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=2977, Available in-common: iperf 
>>>>> nuttcp iperf3
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3002, Using tool: iperf
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3117, Requested Time: 
>>>>> 1409679438.153914
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3119, Latest Acceptable Time:
>>>>> 1409680038.153914
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3292,
>>>>> Reservation(142.104.17.140:4823): 1409679448.061957
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3156, Server '142.104.17.140:4823'
>>>>> accepted test request at time 1409679448.061957
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3292, Reservation(localhost):
>>>>> 1409679448.061957
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3185, Client 'localhost' accepted
>>>>> test request at time 1409679448.061957
>>>>> bwctl[122582]: FILE=bwctl.c, LINE=3359, BWLStartSessions: Failed
>>>>> <server.log><client.log>
>>>>
>>>
>