Index alias confusion
All indices have one or two indices aliases. These aliases help to reduce changes in configurations of the micro services.
Now an issue is, that aliases are used both for live tasks and for creating the next id-hub / linked data indices.
These two should be separated.
-
For this to work properly, we need more storage in elastic search as all indices need to be present twice at the same time. Which is not entirely possible with current disk space (when removing replicas it just so works out). -
Have a workflow management solution which can start all of the parts and remembers what has been done and what has not. And can try failed parts. -
Implement a script for go-live alias moves! -
Make a separation between the Linked Data Workflow and the Id Hub Workflow.
Edited by Sebastian Schüpbach