kwilczynski / packer-templates Goto Github PK

Building from master does not complete when building AMIs for Ubuntu 16.04.

Issue 1:

Error when building ixgbevf:

cd src/; make BUILD_KERNEL=4.4.0-1022-aws....(bad exit status: 2)
ERROR (dkms apport): binary package for ixgbevf: 3.1.2 not found
Error! Bad return status for module build on kernel: 4.4.0-1022-aws (x86_64)
Consult /var/lib/dkms/ixgbevf/3.1.2/build/make.log for more information.

I found a fix in this SO answer.

Adding the following line to /scripts/ec2/networking.sh (I added to line 104) resolves the issue:

sed -i 's/#if UTS_UBUNTU_RELEASE_ABI > 255/#if UTS_UBUNTU_RELEASE_ABI > 99255/' /usr/src/ixgbevf-${SRIOV_DRIVER_VERSION}/src/kcompat.h

Issue 2:

With the above resolved, the AMI builds but when started, #1 is occurring again, this time for /etc/udev/70-persistent-rules.net. cloud-init now calls write_file on this file here which causes the same issue.

Applying the fix in #1 to /scripts/common/clean-up.sh resolves the issue:
-- line 277

if [[ $AMAZON_EC2 == 'yes' ]]; then
    # Get rid of this file, alas clout-init will probably
    # create it again automatically so that it can wreck
    # network configuration. These files, sadly cannot be
    # simply a symbolic links to /dev/null, as cloud-init
    # would change permission of the device node to 0644,
    # which is disastrous, every time during the system
    # startup.
    rm -f \
      /etc/network/interfaces.d/50-cloud-init.cfg \
      /etc/systemd/network/50-cloud-init-eth0.link \
      /etc/udev/rules.d/70-persistent-net.rules

    pushd /etc/network/interfaces.d &>/dev/null
    mknod .null c 1 3
    ln -sf .null 50-cloud-init.cfg
    popd &>/dev/null

    pushd /etc/udev/rules.d &>/dev/null
    mknod .null c 1 3
    ln -sf .null 70-persistent-net.rules
    popd &>/dev/null
fi

If these fixes are correct, I'm happy to submit a PR - just let me know.

And, thank you for all of your work here!

EDIT: Fixed in my fork via fbb0439

After running the ubuntu-16.04-amd64-docker.json and try to spin up a new instance from the AMI created, I'm greeted wit this

`Welcome to Ubuntu 16.04 LTS (GNU/Linux 4.4.0-28-generic x86_64)

              ubuntu-16.04-amd64-docker (amazon-ebs)
                            ___ ___ ___
                           | __/ __|_  )
                           | _| (__ / /
                           |___\___/___|

                   Mon Jul 11 19:15:30 UTC 2016
                          Version: 0.1.0

System information as of Mon Jul 11 21:45:25 UTC 2016

System load: 0.0 Processes: 101
Usage of /: 9.0% of 9.71GB Users logged in: 1
Memory usage: 8% IP address for eth0: 10.0.0.0
Swap usage: 0%

Last login: Mon Jul 11 20:39:48 2016 from xx.xx.xx.xx
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
ubuntu@ip-10-0-0-0:~$ `

where can i track down what script is trying to access /dev/null?

thanks

When running in EC2 generated AMI

docker ps -a or any docker run -d --name=test

just hangs and doesn't give any output or indicators that something is happening

How would you be able to pre-load with docker images?

I notice that if i included my docker pull commands at the end of common/docker.sh that when the artifacts are finished, the docker images are not on the server. But in the output logs they clearly show them being pulled from the docker repositories.

Where would we be able to place the code to have these images be persistent?

thanks.