all the minutes and then some | Continuation: ElasticSearch | | - TC should provide guidance WRT:
- how to use kafka? at present, the mod-search index is populated by mod-inventory-storage directly accessing kafka to publish its updates; may other modules do the same?
- how to use mod-search? at present, mod-search is backed by ES; could it instead by backed by another indexing solution such as lucene? is ES now considered "part of the platform" in a somewhat analogous way to PostgreSQL?
- WRT kafka:
- active discussions/active work is taking place
- TC is not being asked to make a decision about kafka at this time
- WRT mod-search:
- licensing concerns are resolved. ... or should be; TAMU will confirm internally.
- explicitly supporting things with an abstraction layer is too expensive WRT support
- i.e. there must be constraints on implementation options for core modules; while a microservice's public interface is agnostic (HTTP), we can't just pick any language/any service/any etc. and expect full support from dev-ops. it is simply not practically feasible, even if technically feasible.
- Q: What are we saying about mod-search and ES; what is the "point of contact" for an app using ES? must it go through mod-search or can it hit ES directly?
- A: ES, like PostgreSQL, is part of the platform; it is supported, no special approval is approval must be sought from TC or any council/committee to use it. i.e. from a developer's point of view, there are no constraints on the use of ES. apps may access it directly.
|