RARE user and assistance email list

[Frédéric LOUI <>]

> Now it should always work.
Yeah, basically this race condition is invisible with lab having few nodes.

However, when you create a lab with _A_LOT_ nodes this race condition is 
annoying.
It tested this against big labs and now it works 100% without no 
sleep/wait-for hack.

> Le 24 avr. 2023 à 15:48, David Schmitz <> a écrit :
> 
> Hi Frédéric,
> 
> On Mon, 24 Apr 2023, Frédéric LOUI wrote:
> 
>> Date: Mon, 24 Apr 2023 15:39:49 +0200
>> From: Frédéric LOUI <>
>> To: David Schmitz <>
>> Cc: 
>> Subject: Re: ERROR: Permission to rare-freertr/freeRtr-containerlab.git 
>> denied
>>     to dschmitz2017.
>> Hi,
>> 
>> For the race condition please look at this:
>> https://github.com/rare-freertr/freeRtr-containerlab/blob/main/hwdet-init.sh#L18
>> 
>> When launching the lab with containerlab there is a CLAB_INTFS env var 
>> indicating the number of expected interface.
>> 
>> It will help create yours with FoD.
>> You can create a generic container by testing CLAB_INTFS env VAR but in my 
>> case I just decided to create a specific one.
> Thank you.
> I have a similar while loop solution for the single interface eth1 now
> in https://github.com/GEANT/FOD/blob/python3/runfod.sh .
> Now it should always work.
> 
> 
> For freertr which might have a number of interfaces to be injected
> the comparison of the number is an elegant solution.
> I will remember that.
> 
> 
> Best Reards
> David
> 
>> 
>> Frederic
>> 
>>> Le 24 avr. 2023 à 15:32, David Schmitz <> a écrit :
>>> 
>>> Hi Frédéric,
>>> 
>>> obviously the IP address setup of eth1 in cl1 container does not work
>>> reliably yet, because of a race condition.
>>> 
>>> It seems the veth end point is just injected/moved into (like "ip link 
>>> dev eth1 set netns ...")
>>> the container when it is already running.
>>> 
>>> I have to work on a workaround.
>>> 
>>> Best Regards
>>> David
>>> 
>>> On Mon, 24 Apr 2023, David Schmitz wrote:
>>> 
>>>> Date: Mon, 24 Apr 2023 15:04:45 +0200 (CEST)
>>>> From: David Schmitz <>
>>>> To: Frédéric LOUI <>
>>>> Subject: Re: ERROR: Permission to rare-freertr/freeRtr-containerlab.git 
>>>> denied
>>>>    to dschmitz2017.
>>>> Hi Frédéric,
>>>> 
>>>> On Mon, 24 Apr 2023, Frédéric LOUI wrote:
>>>> 
>>>>> Date: Mon, 24 Apr 2023 14:37:52 +0200
>>>>> From: Frédéric LOUI <>
>>>>> To: David Schmitz <>
>>>>> Subject: Re: ERROR: Permission to rare-freertr/freeRtr-containerlab.git 
>>>>> denied
>>>>>    to dschmitz2017.
>>>>> Hi David !
>>>>> If you are OK we can take it at rare-users list.
>>>>> In that case I’ll you cc the list next time !
>>>> Ok.
>>>> 
>>>>> 1- Congrats ! Excellent work !
>>>>> 2- Not sure what you meant by « excluding eth1 » but I trust your 
>>>>> judgement :-)
>>>>>  (Or you might be thinking of containerlab eth0 management address ?)
>>>> eth0 container management IP address is initialized by docker/OS in 
>>>> container (dhcp?),
>>>> and will be changing depending on the other docker containers running on 
>>>> the host
>>>> and can even change when destroying/deploying the clab.
>>>> 
>>>> Of course, that is why we agreed on the extra veth link defined
>>>> in freeRtr-containerlab/lab/005-rare-hello-fod/rtr005.clab.yml
>>>> fod1:eth1 <-> rtr1:eth3 ,
>>>> whose endpoint inside the cl1 container is namend eth1.
>>>> That one is not initialized by the OS (ubuntu in container) nor docker,
>>>> and so the agreed IP address 10.3.10.3 has to be set by some other means.
>>>> In fact, now FoD installer/runtime start script has some simple support 
>>>> for it:
>>>> 
>>>> Check
>>>> inst/testing/fodexabgp-containerlab1/Dockerfile for
>>>> ...
>>>> ... /opt/install-debian.sh ... --ip-addr-set eth1 10.3.10.3 ...
>>>> 
>>>> When the container is started,
>>>> and so /opt/FOD/runfod.sh (CMD) is started inside, that script will also 
>>>> init the eth1 IP address.
>>>> (Of course something like CMD [ "sh" "-c" "ifconfig eth1 10.3.10.3; exec 
>>>> /opt/FOD/runfod.sh" ]
>>>> also would be an option, but that looks more dirty to me,
>>>> especially as the 10.3.10.3 has to match the BGP peer adddress 
>>>> 10.3.10.10,
>>>> which is also specified in the /opt/install-debian.sh ... command line in
>>>> inst/testing/fodexabgp-containerlab1/Dockerfile)
>>>> 
>>>>>> and build the cl1:latest locally:
>>>>> You should be able to build 
>>>>> https://github.com/GEANT/FOD/blob/python3/inst/testing/fodexabgp-containerlab1/Dockerfile
>>>>> Daily for example using GitHub Actions. Seem dig into this during my 
>>>>> spare (vacation) time
>>>>> 3- IIRC there is using docker CMD or other primitives in order to 
>>>>> achieve that.
>>>> I thought so as well,
>>>> but unfortunately seem not to be the case.
>>>> "EXPOSE ..." is only a kind of documentation between provider of a 
>>>> dockerfile
>>>> and someone using that dockerfile:
>>>> https://stackoverflow.com/questions/61234941/docker-expose-and-p .
>>>> Actually, the "-p 8000:8000 " as part of the "docker run ..." is 
>>>> necessary,
>>>> either manually or by containerlab.
>>>> 
>>>>>  But I would suggest you ask to ContainerLab Discord server in #general 
>>>>> channel.
>>>>>  I met the lead developer physically in Paris during MPLS World 
>>>>> Congress and he is super friendly and reactive.
>>>>>  (He is from the Netherlands so European shift)
>>>> 
>>>> Regarding container lab, it supports the following:
>>>> https://containerlab.dev/manual/published-ports/ ,
>>>> but that is too much, it will connect the port to some VPN tunnel system
>>>> to make it available for the public,
>>>> not to make it available on the host.
>>>> 
>>>> Anyway, socat as a workaround for now is ok as well.
>>>> 
>>>>> Keep up the good work !
>>>>> Let’s tackle demo with Eugene/NEMO crew after that :-)
>>>> Ok.
>>>> 
>>>> Best Regards
>>>> David
>>>> 
>>>> 
>>>> Best Regards
>>>> David
>>>> 
>>>>> Frederic
>>>>>> Le 24 avr. 2023 à 12:58, David Schmitz <> a écrit :
>>>>>> Hi Frédéric,
>>>>>> On Mon, 24 Apr 2023, Frédéric LOUI wrote:
>>>>>>> Date: Mon, 24 Apr 2023 11:52:39 +0200
>>>>>>> From: Frédéric LOUI <>
>>>>>>> To: David Schmitz <>
>>>>>>> Subject: Re: ERROR: Permission to 
>>>>>>> rare-freertr/freeRtr-containerlab.git denied
>>>>>>>   to dschmitz2017.
>>>>>>> Corrected.
>>>>>>> You should be able to push now :-)
>>>>>> Thanks. It works.
>>>>>> 1.
>>>>>> Now I have pushed an update with some rtr-sw-conf for rtr1 for 
>>>>>> FlowSpec injection
>>>>>> locally + via BGP from exabgp.
>>>>>> 2.
>>>>>> The container 
>>>>>> https://github.com/GEANT/FOD/blob/python3/inst/testing/fodexabgp-containerlab1/Dockerfile
>>>>>> is slightly updated as well,
>>>>>> in order to include setting of the eth1 interface address when started.
>>>>>> So, the exabgp BGP session should just work.
>>>>>> The next rebuild of 
>>>>>> https://github.com/GEANT/FOD/blob/python3/.github/workflows/docker-publish.yml
>>>>>> referencing the Dockerfile above
>>>>>> should include this.
>>>>>> Anyway, currently 
>>>>>> https://github.com/rare-freertr/freeRtr-containerlab/blob/main/lab/005-rare-hello-fod/rtr005.clab.yml
>>>>>> still references cl1:latest as FoD container.
>>>>>> So, it is still needed to checkout FoD (either main branch "python3" 
>>>>>> or "feature/exabgp_support2")
>>>>>> and build the cl1:latest locally:
>>>>>> (actually only the Dockerfile 
>>>>>> inst/testing/fodexabgp-containerlab1/Dockerfile
>>>>>> and used ./install-debian.sh should be needed, not the whole FoD git 
>>>>>> checkout)
>>>>>> # run from FoD main dir:
>>>>>> docker build -f inst/testing/fodexabgp-containerlab1/Dockerfile -t 
>>>>>> cl1:latest .
>>>>>> 3.
>>>>>> There seems to be no way of making containerlab to map the port 8000 
>>>>>> inside cl1:latest to the outside port on the host
>>>>>> ("docker run .... -p 8000:8000 ....").
>>>>>> Anyway, a workaround for now can be, e.g.
>>>>>> socat TCP-LISTEN:8000,fork TCP-CONNECT:172.20.20.3:8000
>>>>>> (assuming that 172.20.20.3 is the IP address of the management 
>>>>>> interface eth0
>>>>>> of the running cl1:latest container)
>>>>>> Best Regards
>>>>>> David
>>>>>>>> Le 24 avr. 2023 à 08:37, David Schmitz <> a 
>>>>>>>> écrit :
>>>>>>>> HI Frédéric,
>>>>>>>> It seems my last changes on Friday morning fixed the Docker 
>>>>>>>> build/pulish action:
>>>>>>>> https://github.com/GEANT/FOD/actions/runs/4777103267
>>>>>>>> yesterday it run successfully.
>>>>>>>> Best Regards
>>>>>>>> David
>>>>>>>> On Mon, 24 Apr 2023, David Schmitz wrote:
>>>>>>>>> Date: Mon, 24 Apr 2023 08:18:37 +0200 (CEST)
>>>>>>>>> From: David Schmitz <>
>>>>>>>>> To: Fréderic LOUI <>
>>>>>>>>> Subject: ERROR: Permission to rare-freertr/freeRtr-containerlab.git 
>>>>>>>>> denied to
>>>>>>>>> dschmitz2017.
>>>>>>>>> Hi Frédéric,
>>>>>>>>> I tried to push my rtr config for rtr1 in containerlab
>>>>>>>>> (basically, file 
>>>>>>>>> ./lab/005-rare-hello-fod/clab-rtr005/rtr1/run/conf/rtr-sw.txt).
>>>>>>>>> But it failed:
>>>>>>>>> ERROR: Permission to rare-freertr/freeRtr-containerlab.git denied 
>>>>>>>>> to dschmitz2017.
>>>>>>>>> fatal: Could not read from remote repository.
>>>>>>>>> Could you grant user dschmitz2017 write/push access?
>>>>>>>>> Best Regards
>>>>>>>>> David
>>>>>>>> --
>>>>>>>> David Schmitz
>>>>>>>> Boltzmannstrasse 1, 85748 Garching
>>>>>>>> Telefon:   +49 89 35831-8765
>>>>>>>> Leibniz-Rechenzentrum, Germany
>>>>>>>> Mail:  
>>>>>> --
>>>>>> David Schmitz
>>>>>> Boltzmannstrasse 1, 85748 Garching
>>>>>> Telefon:   +49 89 35831-8765
>>>>>> Leibniz-Rechenzentrum, Germany
>>>>>> Mail:  
>>>> 
>>>> 
>>> 
>>> --
>>> 
>>> David Schmitz
>>> 
>>> Boltzmannstrasse 1, 85748 Garching
>>> Telefon:   +49 89 35831-8765
>>> Leibniz-Rechenzentrum, Germany
>>> Mail:  
>>> 
>>> 
>> 
> 
> -- 
> 
> David Schmitz
> 
> Boltzmannstrasse 1, 85748 Garching
> Telefon:   +49 89 35831-8765
> Leibniz-Rechenzentrum, Germany
> Mail:  
> 
>