[cip-dev] Problems installing "Board at desk"

[robert.marshall@codethink.co.uk (Robert Marshall)]

Daniel

Thanks for this - comments below...

"Daniel Sangorrin" <daniel.sangorrin@toshiba.co.jp> writes:

> Hi,
>
> I am trying the "board-at-desk-single-dev" project (sorry to do it so late).
> I managed to build the CIP kernel and see the job results using KernelCI.
> However, I am having problems with LAVA health checks (see at the end).
>
> First, I would like to report on a few problems I had to solve to get the kernel built:
>
> *************************************************
> 1) Although now I know that the current development occurs at
> https://gitlab.com/cip-project/board-at-desk-single-dev.git
> googling "CIP kernelci" gives also the following outdated (?) sites which can 
> be confusing:
> https://github.com/cip-project/cip-kernelci.git
> https://gitlab.com/cip-project/kernelci-debian.git
>
> Q: are they necessary? or is it some misunderstanding from my side?
>

I think this is largely addressed by Wolfgang's comment

> 2) Problems behind a proxy
>
> a) I did the following to setup proxy settings for vagrant on Ubuntu 16.04 Xenial:
>
>    $ sudo apt-get remove vagrant <-- gives errors when installing vagrant-proxyconf
>    $ dpkg -i vagrant_1.9.4_x86_64.deb
>    $ vagrant plugin install vagrant-proxyconf
>    $ vi Vagrantfile
> + if Vagrant.has_plugin?("vagrant-proxyconf")
> +   config.proxy.http     = "http://xxx:yyyy/"
> +   config.proxy.https    = "https://xxx:yyy/"
> +   config.proxy.no_proxy = "127.0.0.1,localhost,xxxx."
> + end
>
> Q: maybe it would be good to add this to the tutorial

The tutorial recommends 1.8.1 but that's the default version for 16.04
so any tutorial change needs to address both those concerns

>
> b) I got an error during vagrant up
>
> ==> default: fatal: [kernel-ci-backend]: FAILED! => {"changed": false, "cmd": "/usr/bin/apt-key adv --keyserver hkp://keyserver.ubuntu.com --recv EA312927", "failed": true, 
> - Solved it by adding port 80 for apt-key
>    $ vagrant ssh
>     guest$ vi kernelci-backend/roles/install-deps/tasks/install-mongodb.yml
>                 - hkp://keyserver.ubuntu.com
> 	+hkp://keyserver.ubuntu.com:80
>      [Alt]  sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv EA312927
>
> Q: if that works without proxies, maybe it should be set to 80 by default?
>

Will test and see, if so there'll probably need to be a change in install_backend.sh

> c) I got a lot of warnings like these ones
>
> - Warning 1 (ignore)
>     GetPassWarning: Can not control echo on the terminal.? or ?Warning: Password input may be echoed.? - These do not affect the operation of the KernelCI VM.
> - Warning 2 (ignore)
>     ==> default: lava_scheduler_app.Notification.job_status_trigger: (fields.W901) CommaSeparatedIntegerField has been deprecated. Support for it (except in historical migrations) will be removed in Django 2.0.
>     ==> default: 	HINT: Use CharField(validators=[validate_comma_separated_integer_list]) instead.
>
> Q: The tutorial mentions Warning 1, but not Warning 2. Maybe adding that would be a good idea. 

The tutorial is going to be reorganised before the release so we will
take this on board.

>
> 3) Modifying the 8080 port (very commonly used port, e.g. Fuego ;_+)
>
> I solved this by
>    $   vi Vagrantfile
>        +  config.vm.network :forwarded_port, guest: 8081, host: 8081
>     $ sudo vi /etc/apache2/ports.conf
>         -> change to 8081
>     $ sudo vi /etc/apache2/sites-enabled/lava-server.conf
>         -> change to 8081
>     $ sudo service apache2 restart
>     $ sudo /etc/init.d/lava-server restart
>
> Q: maybe this could be automated (?)

This would be the right solution - for the moment the best approach is
to edit integration-scripts/install_lava.sh on the host before creating
the VM and change 8080 to the desired port there (as well as the
Vagrantfile mod)

> *************************************************
>
> Second, regarding to LAVA health checks I think this is again a problem with being behind a
> proxy but I'm not sure how to debug it. These are the error messages that I get
> with QEMU's health check (/vagrant/tests/qemu-health-check.yaml)
>
>     - log:
>         Root tmp directory created at /var/lib/lava/dispatcher/tmp/7
>         start: 0 validate
>         Validating that https://images.validation.linaro.org/kvm/standard/stretch-2.img.gz exists
>         no device environment specified
>         Invalid job definition
>         Invalid job data: ["HTTPSConnectionPool(host='images.validation.linaro.org', port=443): Max retries exceeded with url: /kvm/standard/stretch-2.img.gz (Caused by NewConnectionError('<requests.packages.urllib3.connection.VerifiedHTTPSConnection object@0x7f430c9b9c50>: Failed to establish a new connection: [Errno -5] No address associated with hostname',))"]
>         validate duration: 0.02
>         Cleanup: removing /var/lib/lava/dispatcher/tmp/7
>     - traceback
>         Traceback (most recent call last):
>           File "/usr/lib/python2.7/dist-packages/lava/dispatcher/commands.py", line 88, in run_pipeline_job
>             job.validate(simulate=validate_only)
>           File "/usr/lib/python2.7/dist-packages/lava_dispatcher/pipeline/job.py", line 173, in validate
>             self.pipeline.validate_actions()
>           File "/usr/lib/python2.7/dist-packages/lava_dispatcher/pipeline/action.py", line 205, in validate_actions
>             raise JobError("Invalid job data: %s\n" % self.errors)
>         JobError: Invalid job data: ["HTTPSConnectionPool(host='images.validation.linaro.org', port=443): Max retries exceeded with url: /kvm/standard/stretch-2.img.gz (Caused by NewConnectionError('<requests.packages.urllib3.connection.VerifiedHTTPSConnection object@0x7f430c9b9c50>: Failed to establish a new connection: [Errno -5] No address associated with hostname',))"]


Looking at this at the moment

>
> If someone has a clue about this please let me know. 
> # http_proxy-like variables are all defined in the VM (/etc/environment), and I can use wget and download
> stretch-2.img.gz without problems.
>

Thanks for your report and suggestions

Robert