docs: expand the Milvus migration warning

CHANGELOG.md

@@ -14,7 +14,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - 🧲 **Reordering filtered model lists.** Models on the admin models page can now be dragged into place while a search, view or tag filter is active, with every model hidden by the filter keeping its position. [#30390](https://github.com/open-webui/open-webui/pull/30390), [#29634](https://github.com/open-webui/open-webui/issues/29634)
 - 🖨️ **Images in note PDFs.** Downloading a note as a PDF now includes the images pasted into it, where the PDF held only the title and text. [#30975](https://github.com/open-webui/open-webui/pull/30975), [#30974](https://github.com/open-webui/open-webui/issues/30974)
 - 🕵️ **Tavily search depth.** Administrators can now choose how deep Tavily web searches go, from ultra-fast to advanced, in the Tavily settings of the web search page or through "TAVILY_SEARCH_DEPTH", separately from the existing extract depth. [#31308](https://github.com/open-webui/open-webui/pull/31308), [#29891](https://github.com/open-webui/open-webui/issues/29891)
-- 🏎️ **Native hybrid search on Milvus.** Hybrid search on Milvus 2.5 and newer, with one collection per knowledge base or with multitenancy, now runs the keyword half inside Milvus with its built-in BM25 full-text search and merges it with the vector results, where every search fetched every chunk of the collection and scored it in Open WebUI; existing collections get a text-only companion collection built once at startup, and older Milvus servers keep the previous hybrid search. [#31645](https://github.com/open-webui/open-webui/pull/31645), [#26243](https://github.com/open-webui/open-webui/issues/26243)
+- 🏎️ **Native hybrid search on Milvus.** Hybrid search on Milvus 2.5 and newer, with one collection per knowledge base or with multitenancy, now runs the keyword half inside Milvus with its built-in BM25 full-text search and merges it with the vector results, where every search loaded every chunk of the collection into Open WebUI and scored it there, making hybrid search on Milvus much faster and lighter on memory; existing collections get a text-only companion collection built once at startup, and older Milvus servers keep the previous hybrid search. [#31645](https://github.com/open-webui/open-webui/pull/31645), [#26243](https://github.com/open-webui/open-webui/issues/26243)
 - 🪙 **Input and output tokens in analytics.** Hovering, focusing or clicking the token total on the admin Analytics dashboard now shows how many of those tokens were input and how many were output. [Commit](https://github.com/open-webui/open-webui/commit/7e317fbada4acc5c48f326aa5935b3928275ea78), [#31646](https://github.com/open-webui/open-webui/issues/31646)
 - 🚀 **Faster JSON handling across the app.** Every instance now reads and writes JSON with orjson, speeding up request and response bodies, streamed provider responses and live socket updates, which now encode about 17 times and decode about 3 times faster. [#31616](https://github.com/open-webui/open-webui/pull/31616)
 - 🔄 **General improvements.** Various improvements were implemented across the application to enhance performance, stability, and security.
@@ -173,7 +173,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
-- ⚠️ **IMPORTANT for Milvus and Milvus multitenancy users: back up your Milvus data before upgrading**: This release migrates every existing Milvus collection on its first start with "ENABLE_DB_MIGRATIONS" on (the default) by building a text-only BM25 companion collection next to it and copying all chunk text into it, and Open WebUI only finishes starting once that copy is done. On large databases this can take a long time, around 40 minutes for 200 GB with 1536-dimension embeddings on a single standalone Milvus server, and it needs extra storage roughly the size of the stored chunk text. We strongly recommend backing up your Milvus data before upgrading, planning for a much longer first startup and not interrupting it; the original vectors and indexes are left untouched, and collections that cannot be copied and Milvus servers older than 2.5 keep using the previous hybrid search. [#31645](https://github.com/open-webui/open-webui/pull/31645)
+- ⚠️ **IMPORTANT for Milvus and Milvus multitenancy users: back up your Milvus data before upgrading**: On its first start with "ENABLE_DB_MIGRATIONS" on (the default), this release copies the chunk text of every existing Milvus collection into a new BM25 companion collection, and Open WebUI only finishes starting once that is done, which took 40 to 45 minutes for 200 GB with 1536-dimension embeddings on one standalone Milvus server, so plan for a long first startup and enough free disk space for a second copy of the chunk text. Nothing is re-embedded and the original collections are never changed or deleted: a failed copy is retried with smaller batches and again on the next start, and the new collection is only used once its copy has fully succeeded. The migration was tested extensively on real installations, including across a Milvus 2.6 to 3.0.2 upgrade, but no testing covers every setup, so back up first and avoid upgrading Milvus and Open WebUI on the same day where you can. Native hybrid search needs Milvus 2.5 or newer; older servers skip the migration and keep the previous hybrid search. [#31645](https://github.com/open-webui/open-webui/pull/31645)
 - ⚡ **orjson on by default.** "ENABLE_ORJSON" now defaults to on, and setting "ENABLE_ORJSON=false" switches back to the standard JSON encoder. [#31616](https://github.com/open-webui/open-webui/pull/31616)
 - 🔁 **Update Redis-backed instances together.** Where several servers share their websocket traffic through Redis, every instance should be updated at the same time, since live chat updates sent by an updated instance do not reach one still on an older version unless "WEBSOCKET_REDIS_ROOM_CHANNELS" is set to false. [#28818](https://github.com/open-webui/open-webui/pull/28818)