(CAT-2052) Pass target container URI instead of container SHA ID to add_feature_to_node() method #574
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #574 +/- ##
==========================================
- Coverage 64.03% 63.86% -0.17%
==========================================
Files 6 6
Lines 759 761 +2
==========================================
Hits 486 486
- Misses 273 275 +2 ☔ View full report in Codecov by Sentry. |
…dd_feature_to_node() method
There is a bug in the 'install_agent' task.
It is supposed to install the agent on a host (a VM or a docker container) using bolt and then add the 'puppet-agent' feature to the host in the litmus_inventory file.
Bolt returns the SHA ID of the container instead of the localhost URI after it installs the agent.
The feature is added through the add_feature_to_node() method. The last parameter of the method is being sent as result["target"]. This is fine in the case of VMs since its their IPv4 address and the method is expecting just that.
But in case of docker containers it is their SHA container ID and the method expecting the localhost:{port} URI. This results in the feature not being added to the host.
Since bolt is returning the results in the same order it is provided the input 'targets', we can add the feature by indexing the 'targets' array in that order.
See https://perforce.atlassian.net/browse/CAT-2052?focusedCommentId=2970824
shubhamshinde360
force-pushed
the
CAT-2052-fix-install-agent-bug
branch
from
8a020e5
to
63424d1
Compare
shubhamshinde360
changed the title
|
When I use this change and then run the nightly tests: https://github.com/puppetlabs/puppetlabs-package/actions/runs/11112007788 |
There was a problem hiding this comment.
verified with docker_exp, docker, winrm and ssh nodes. Nice catch @shubhamshinde360, thanks again!
Summary
There is a bug in the 'install_agent' task.
It is supposed to install the agent on a host (a VM or a docker container) using bolt and then add the 'puppet-agent' feature to the host in the litmus_inventory file.
Bolt returns the SHA ID of the container instead of the localhost URI after it installs the agent. The feature is added through the add_feature_to_node() method.
The last parameter of the method is being sent as result["target"]. This is fine in the case of VMs since its their IPv4 address and the method is expecting just that.
But in case of docker containers it is their SHA container ID and the method expecting the localhost:{port} URI. This results in the feature not being added to the host.
Since bolt is returning the results in the same order it is provided the input 'targets', we can add the feature by indexing the 'targets' array in that order.
Additional Context
See https://perforce.atlassian.net/browse/CAT-2052?focusedCommentId=2970824
The puppetlabs-package nightlies are failing due to this on all the docker provisioned VMs.
Related Issues (if any)
Mention any related issues or pull requests.
Checklist