The dependency links are not created in the jaeger-collector but in an external process. This process aggregates spans and calculates which links should be created. It directly accesses the storage.
So to support dep. links we would have to add support for clickhouse to spark-dependencies. Or maybe there is a way to implement it directly in the storage (e.g. via view) but I cannot confirm that.
This is more tricky. Let me explain why.
The dependency links are not created in the jaeger-collector but in an external process. This process aggregates spans and calculates which links should be created. It directly accesses the storage.
The implementation is here https://github.com/jaegertracing/spark-dependencies
So to support dep. links we would have to add support for clickhouse to spark-dependencies. Or maybe there is a way to implement it directly in the storage (e.g. via view) but I cannot confirm that.