The indexing pipeline (T422179) will prepare the necessary triple-store index files for a WDQS pod (T425007) to be initiated.
The idea is to have an init container which performs the following steps:
- Check if the index as prepared by the pipeline is newer than the one in the pods node storage (or none exists on the node)
- If so, update the index:
- Move the current index files to a backup location
- Copy the new index files from the pipeline result to the pod node
- Update the Kafka offset for the node based on the index dump date
- (!) Write the Kafka group for the node into a file to be read by the consumer (It is still unclear how we solve the issue of the Kafka group per node)
AC:
- A new docker image for the init container is available in the docker registry
- The init container has been integrated into the helm chart (T425007) (draft enough)