Skip to content

Latest commit

 

History

History
204 lines (156 loc) · 10.7 KB

README.md

File metadata and controls

204 lines (156 loc) · 10.7 KB

CATalogASB Local Deployment

Overview

These playbooks will do the following in a local environment:

Pre-Reqs

  • Ansible 2.4.0+ installed.
  • Docker installed and configured

    • Suggestion, to ease usage we allow our regular user to access the docker server by doing the below:

      • sudo groupadd docker
      • sudo usermod -aG docker $USER
      • sudo systemctl restart docker

    • Follow the Docker setup instructions from oc cluster up documentation to configure the necessary Docker networking settings. Stop before installing the oc binary.

Execute

  • Copy config/my_vars.yml.example to config/my_vars.yml and edit as needed. You can use the my_vars.yml to override any settings. For example: 
    $ cp config/my_vars.yml.example config/my_vars.yml
$ vim config/my_vars.yml

    • If using a private repository, set dockerhub_user_name (and optionally dockerhub_user_password) with your own dockerhub username (and password). This will skip the prompts during execution and makes re-runs easy. A valid dockerhub login is required for the broker to authenticate to dockerhub to search an organization for APBs.

    • Set dockerhub_org to load APB images into your broker from an organization. For dockerhub organization you may use your own if you pushed your own APBs or you may use the ansibleplaybookbundle organization as a sample.

    • Set hostname and openshift_routing_suffix if you want to use a different static IP.

    • Example my_vars.yml $ cat local/config/my_vars.yml ---

      dockerhub_user_name: [email protected]
# dockerhub_user_password: changeme  # if commented out, will prompt
dockerhub_org: ansibleplaybookbundle
  • Navigate to the local/linux folder and run the script to set up OpenShift. 
    $ cd local/linux
$ ./run_setup_local.sh
  • Open a Web Browser
    • Visit: https://apiserver-service-catalog.CLUSTERIP.nip.io
      • Accept the SSL certificate for the apiserver-service-catalog endpoint
      • Ignore the text that appears and proceed to the main OpenShift URL next
      • Note: must accept the new SSL cert, each time you reset your OpenShift environment
    • Visit: https://<CLUSTERIP>.nip.io:8443
      • The is the same as the one you set in common_vars

Cleanup

  • To terminate the local instance run the below

    $ oc cluster down

  • To reset the environment to a clean instance of origin with ASB and Service Catalog run the below

    $ ./reset_environment.sh

Testing downstream images

  • Use the --rcm flag. For instance:
    • ./run_setup_local.sh --rcm
    • ./reset_environment.sh --rcm

Tested with

  • ansible 2.4.1.0
    • Problems were seen using ansible 2.2 and lower

Troubleshooting

pull() got an unexpected keyword argument 'decode'

Error pulling image docker.io/ansibleplaybookbundle/ansible-service-broker-apb:summit - pull() got an unexpected keyword argument 'decode'

This is a problem with having docker-py installed, and at a specific version. More info in ansible/ansible-modules-core#5515. The recommended fix for this is to uninstall docker-py, which needs to be uninstalled: For EL7: sudo yum remove python-docker-py For Fedora: sudo dnf remove python-docker-py For Others: sudo pip uninstall docker-py

APBs not visible from OpenShift Web UI

In some cases APBs won't be visible from the OpenShift Console after ./run_setup_local.sh. This can happen when the catalog is unable to talk to the broker due to an issue with iptables.

The recommended fix is to flush iptables rules and reset the catasb environment.

sudo iptables -F
./reset_environment.sh

Cannot connect to the Docker daemon. Is the docker daemon running on this host

TASK [openshift_setup : Resetting cluster, True]
 ****************
"error: cannot communicate with Docker"

This may be a permission issue, we recommend relaxing permissions on docker so your regular user is able to access docker. The following will allow your regular user to access docker:

  • sudo groupadd docker
  • sudo usermod -aG docker $USER
  • sudo systemctl restart docker

sudo: a password is required

TASK [openshift_setup : Resetting cluster, True] *********************************************************************************************************************************************************
changed: [localhost]

TASK [openshift_setup : Install docker through pip as it's a requirement of ansible docker module] *******************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "failed": true, "module_stderr": "sudo: a password is required\n", "module_stdout": "", "msg": "MODULE FAILURE", "rc": 1}
	to retry, use: --limit @/home/tsanders/Workspace/catasb/ansible/setup_local_environment.retry
TASK [openshift_setup : Resetting cluster, True] *********************************************************************************************************************************************************
changed: [localhost]

TASK [openshift_setup : Install docker through pip as it's a requirement of ansible docker module] *******************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "failed": true, "module_stderr": "sudo: a password is required\n", "module_stdout": "", "msg": "MODULE FAILURE", "rc": 1}
	to retry, use: --limit @/home/tsanders/Workspace/catasb/ansible/setup_local_environment.retry

We currently run with NOPASSWD configured for sudo, to do the same:

  • sudo visudo

  • Add this line:

    %wheel  ALL=(ALL)       NOPASSWD: ALL

Error: did not detect an --insecure-registry argument on the Docker daemon

TASK [openshift_setup : Run oc cluster up] ***************************************************************************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": true, "cmd": "/home/tsanders/bin/oc cluster up --routing-suffix=172.17.0.1.nip.io --public-hostname=172.17.0.1 --tag=latest --host-config-dir=/var/lib/origin/openshift.local.config", "delta": "0:00:00.136821", "end": "2017-06-26 10:32:31.792848", "failed": true, "rc": 1, "start": "2017-06-26 10:32:31.656027", "stderr": "", "stderr_lines": [], "stdout": "Starting OpenShift using openshift/origin:latest ...\n-- Checking OpenShift client ... OK\n-- Checking Docker client ... OK\n-- Checking Docker version ... OK\n-- Checking for existing OpenShift container ... OK\n-- Checking for openshift/origin:latest image ... OK\n-- Checking Docker daemon configuration ... FAIL\n   Error: did not detect an --insecure-registry argument on the Docker daemon\n   Solution:\n\n     Ensure that the Docker daemon is running with the following argument:\n     \t--insecure-registry 172.30.0.0/16", "stdout_lines": ["Starting OpenShift using openshift/origin:latest ...", "-- Checking OpenShift client ... OK", "-- Checking Docker client ... OK", "-- Checking Docker version ... OK", "-- Checking for existing OpenShift container ... OK", "-- Checking for openshift/origin:latest image ... OK", "-- Checking Docker daemon configuration ... FAIL", "   Error: did not detect an --insecure-registry argument on the Docker daemon", "   Solution:", "", "     Ensure that the Docker daemon is running with the following argument:", "     \t--insecure-registry 172.30.0.0/16"]}
	to retry, use: --limit @/home/tsanders/Workspace/catasb/ansible/setup_local_environment.retry

PLAY RECAP ***********************************************************************************************************************************************************************************************
localhost                  : ok=25   changed=4    unreachable=0    failed=1

There are several configurations required to run oc cluster up, please be sure to read and follow the oc cluster up documentation here: https://github.com/openshift/origin/blob/master/docs/cluster_up_down.md#linux

oc cluster up hangs while starting the cluster

Sometimes catasb will get stuck at starting the cluster phase. It will simply sit at this prompt for what seems like years:

TASK [openshift_setup : Run oc cluster up to start the cluster] ***

This usually means oc cluster up is failing. There are a number of things you need to look at.

  • get log output from a stock oc cluster up.
oc cluster up --service-catalog=true --loglevel=10

If this works take note of the origin version it pulled down. You might want to use that tag in your my_vars.yml.

  • look at your journalctl -e output

Any reference to eviction, usually indicates you are in need of more disk space.

Jan 10 09:41:08 speed3 dockerd-current[21270]: W0110 14:41:08.646423   23589 eviction_manager.go:332] eviction manager: attempting to reclaim imagefs
Jan 10 09:41:08 speed3 dockerd-current[21270]: I0110 14:41:08.646459   23589 helpers.go:1070] eviction manager: attempting to delete unused containers
Jan 10 09:41:08 speed3 dockerd-current[21270]: I0110 14:41:08.651090   23589 helpers.go:1080] eviction manager: attempting to delete unused images

If your usage percentage is more than 85%, you need to free up more disk space. My usage was up to 90% with 49G free. That caused kube to evict pods that were needed to start origin cluster.

$ df -h
Filesystem              Type      Size  Used Avail Use% Mounted on
devtmpfs                devtmpfs  9.6G     0  9.6G   0% /dev
tmpfs                   tmpfs     9.6G  100M  9.5G   2% /dev/shm
tmpfs                   tmpfs     9.6G  2.3M  9.6G   1% /run
tmpfs                   tmpfs     9.6G     0  9.6G   0% /sys/fs/cgroup
# exceeds the threshold of 85%
/dev/mapper/fedora-root ext4      460G  388G   49G  89% /
tmpfs                   tmpfs     9.6G  258M  9.3G   3% /tmp
/dev/sda1               ext4      477M  184M  264M  42% /boot
tmpfs                   tmpfs     2.0G   20K  2.0G   1% /run/user/4
tmpfs                   tmpfs     2.0G  124K  2.0G   1% /run/user/1000

As can be seen in this origin bug comment:

openshift/origin#18046 (comment)

Free up space until you are below the 85%, I went down to 61%. Cluster comes up fine.