4 Commits (44e8422dc468e8071f6e68b7cd651f45abaf30e9)

Author	SHA1	Message	Date
Karsten Jeschkies	44e8422dc4	Enalbe multi tenant queries on Helm test chart. (#8228) ... Co-authored-by: Christian Haudum <christian.haudum@gmail.com>	2 years ago
Trevor Whitney	eb39e2645f	[helm] allow provisioned tokens to be created in different namespaces (#8081) ... **What this PR does / why we need it**: _Note_: This PR bumps the chart to 4.0, so it includes a few other breaking changes we've been waiting to make. The new helm chart has a few different ways to create tokens when running in enterprise mode: * The tokengen job creates the initial admin token * The provisioner job creates the self-monitoring tenant if enabled * The provisioner job can also create additional tenants if configured The problem with these jobs is they put the generated read/write tokens for these tenants in secrets in the same namespace as the loki deploiyment. This is problematic if you need these tokens for a specific application, for example a Grafana datasource, that is deployed in a separate namespace. This PR introduces two breaking changes, one of which is breaking as it significantly changes how both the self monitoring and additional tenants are configured. First it adds the value `enterprise.adminToken.additionalNamespaces` and moves `enterprise.adminTokenSecret` to `enterprise.adminToken.secret`. If `additionalNamespaces` are included, additional secrets containing the GEL admin token are created in those namespaces as well as the release namespace. Second, it changes the structure of defining a tenant from just a string name, to an object containing a `name` and `secretNamespace` field. The `secretNamespace` is the namespace the secret containing that tenants token(s) will be placed in. In the case of the self-monitoring token, a secret will still be created in the default namespace in addition to the `secretNamespace` specified as that token is required by multiple components.	2 years ago
Trevor Whitney	3f9b9cc961	[helm] fix race between minio bucket job and enterprise tokengen job (#8064) ... **What this PR does / why we need it**: This PR replaces https://github.com/grafana/loki/pull/7985. It solves the same problem, which is the race condition between the minio create bucket job and the enterprise tokengen job, but without removing minio.	2 years ago
Trevor Whitney	98fd01aef8	Add empty k3d cluster for testing helm chart (#7984) ... **What this PR does / why we need it**: Add a new k3d cluster which is a cluster that has all Loki dependencies for purposes of testing the helm charts using `helm install`. The other environments use the helm charts, but deploy them using `tanka` and therefore can't catch certain race conditions or issues that occur with helm installs.	2 years ago