-
Notifications
You must be signed in to change notification settings - Fork 35
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Introduce Gnocchi helm chart to Genestack #109
Conversation
Not ready for merge just yet, will be following up with one or two more commits to add documentation where appropriate regarding installing postgresql and gnocchi. |
408224c
to
35ad775
Compare
Getting a 403 from the Gnocchi API pod over its service port 8041 -- looks like apache is denying requests due to...
That path doesn't exist because.... we have python3.x now. There is a place to update the apache configuration in the |
Past the previous issue, now getting a 500 response. There's a traceback in the gnocchi api logs related to there not being a section for gnocchi+keystone within |
Getting a 200 finally:
Note: Looking at the version response, the build is 4.4.4. The gnocchi_api image being used is quay.io/openstack.kolla/ubuntu-source-gnocchi-api:yoga - built only about a month ago. Gnocchi is up to 4.6.1, so this is pretty far behind.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A couple nits, and one general question.
Additionally, it looks like we're following openstack-helm's deployment of postgres, could we use the upstream postgres operator? https://postgres-operator.readthedocs.io/en/latest/
This chart adds PostgreSQL. It's only planned use case at the moment is for Gnocchi indexer storage. A few important notes about this chart are below for consideration. - MySQL should be a supported driver for the Gnocchi indexer but I was unable to get that working. Considering Gnocchi cites preferring PostgreSQL as an indexer, and it's the default in the chart, opting for that at the moment in lieu of fighting the chart to do something it might need major overhauls to accomplish. - There is a an upstream operator for PG we should look into mid to long-term. This chart is only needed to make Gnocchi go, so that we can integrate Ceilometer.
This chart begins to onboard Gnocchi, the Openstack Metrics API. For now we are using the PostgreSQL driver for indexer storage, as it's unclear how to change it over to MySQL. Moreover, Gnocchi recommends using PostgreSQL anyway. Should it be determined later that we can safely use our existing MariaDB implementation in Genestack with no major drawbacks, we should migrate over. This chart is not well maintained, requires updated images, and several patches to make it work with Genestack.
Most in the helm chart were many years old; py2 was being used.
Needed to alter the storage-init script to correct now deprecated syntax for ceph auth capabilities for a given user. The original script uses `profile` in lieu of `allow` in several places.
There's a dependency on a mount related to this secret in several places. It's use is as a fallback value if another is not provided. Provide this keyring secret to avoid helm chart deployment failure.
We upgraded to python3 by pulling in a newer image, consequently this path needed to be updated.
These may need further customization depending on the environment gnocchi will be run. Aside from the pipeline:main section, this is the default configuration in the latest version of gnocchi.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
This PR introduces helm configs for Gnocchi (OpenStack Metrics API) and its dependency, PostgreSQL. Gnocchi is a dependency of Ceilometer, which we will be adding to Genestack soon.
Note that PostgreSQL is being used as the indexer storage back-end in favor of our existing MariaDB operator implementation for several reasons.
Iff we go with PostgreSQL long term, we will want to investigate the upstream operator. The existing chart works and is being maintained to some extent, but the operator no doubt has more active contributors and will likely suit Genestack better.