Update Jan 7 2025: the intention of this issue is to do away with awareness in the code of the differences between primary and archive storage. The two implementations are generally identical, have identical reader/writer API, with only minor changes in behavior that can be managed with more explicit parameters instead if a hardcoded isArchive flag. For example, in the "legacy" mode the ES primary storage uses manual rotation of indices, e.g. by creating a new index name every rollover period, whereas archive storage always uses a fixed index name. The latter behavior can be reproduced by setting use_aliases flag instead of having an explicit understanding of isArchive mode.

This unification becomes especially important in Jaeger-v2 where storage backends are already declared uniformly in the jaeger_storage extension without any knowledge of primary/archive distinction. The distinction only appears once jaegerquery extension retrieves a specific backend by name and uses it as either primary or archive backend, by which point the backend itself is already initialized. The intent is to expose required configuration options to the user such that they can customize the behavior of the backend used for archive as they see fit, e.g. if they still want to rotate the ES indices, they can do so.

This thread #6060 (comment)