github-actions[bot] commented on code in PR #67918:
URL: https://github.com/apache/doris/pull/67918#discussion_r4047556030
##########
be/src/storage/index/inverted/inverted_index_parser.cpp:
##########
@@ -211,62 +216,90 @@ std::string get_analyzer_name_from_properties(
}
std::string normalize_analyzer_key(std::string_view analyzer) {
Review Comment:
[P1] Preserve old-FE reader keys during BE-first upgrades
Doris upgrades BEs before FE, but an old FE lowercases a saved mixed-case
custom analyzer (for example, `Foo`) to `analyzer_name=foo`. The old BE also
normalized the physical reader's `analyzer=Foo` to `foo`; after this change the
new BE registers it only under exact `Foo`. Because the selector refuses to
fall back for a nonempty key, MATCH now returns `INVERTED_INDEX_BYPASS` even
when this is the only reader and silently scans rows until the FE is upgraded.
Such exact metadata predates index-property canonicalization and is one of the
migration states this PR preserves. Please retain a collision-safe normalized
alias/fallback for old-FE requests, or gate exact-key semantics on a
capability/version boundary, and cover the base-FE Thrift shape against the new
selector.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
