* Fix probes when Qdrant is served over TLS
This also improves the integration test runtime and reliability by not installing a fresh cluster for every test but upgrading it with the new config.
Fixes https://github.com/qdrant/qdrant-helm/issues/77
* Deactivate liveness and startup probes by default
Since these probes restart the pod on failure, they can disrupt the health of the Qdrant cluster unnecessarily and cause more problems then actually help. Having a readinessProbe is usually enough.
* Fix tests
* Try to fix connection issues
* Make the qdrant api key easily configurable
This adds a new `apiKey` value.
If `true` it generates a random apiKey, stores it in a secret and ensures that it stays the same during upgrades.
If a string is passed, this string is used as the api key.
The secret is mounted as a `local.yaml` config into the container. See also https://qdrant.tech/documentation/guides/configuration/#order-and-priority.
Alternatively, there are two new values `additionalVolumes` and `additionalVolumeMounts` which allow to mount additional existing volumes, such as pre-existing secrets into the qdrant container.
This also adds bats-core based integration tests to test the correct api key handling
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
* Fix bats integration tests
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
* Fix bats integration tests
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
* Fix bats integration tests
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
* Fix bats integration tests
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
* Fix statefulset
* doc: fix typo
---------
Signed-off-by: Bastian Hofmann <mail@bastianhofmann.de>
Co-authored-by: Tim Eggert <tim@elbart.com>
