- Adds a flag for `compressResponse` and `flushFix` to the request body to be able to overrule compression settings inferred from headers.
- Updates the developer examples with a toggle to run requests with compression enabled or disabled.
- Adds support for backpressure handling for response streams.
- The backpressure update includes a fix where uncompressed streams would never start streaming to the client.
- The analysis endpoint for Explain Log Rate Spikes now includes a ping every 10 seconds to keep the stream alive.
- Integration tests were updated to test both uncompressed and compressed streaming.
Builds out UI/code boilerplate necessary before we start implementing the feature's own UI on a dedicated page.
- Updates navigation to bring up data view/saved search selection before moving on to the explain log spike rates page.
The bar chart race demo page was moved to the aiops/single_endpoint_streaming_demo url. It is kept in this PR so we have two different pages + API endpoints that use streaming. With this still in place it's easier to update the streaming code to be more generic and reusable.
- The url/page aiops/explain_log_rate_spikes has been added with some dummy request that slowly streams a data view's fields to the client. This page will host the actual UI to be brought over from the PoC in follow ups to this PR.
- The structure to embed aiops plugin pages in the ml plugin has been simplified. Instead of a lot of custom code to load the components at runtime in the aiops plugin itself, this now uses React lazy loading with Suspense, similar to how we load Vega charts in other places. We no longer initialize the aiops client side code during startup of the plugin itself and augment it, instead we statically import components and pass on props/contexts from the ml plugin.
- The code to handle streaming chunks on the client side in stream_fetch.ts/use_stream_fetch_reducer.ts has been improved to make better use of TS generics so for a given API endpoint it's able to return the appropriate coresponding return data type and only allows to use the supported reducer actions for that endpoint. Buffering client side actions has been tweaked to handle state updates more quickly if updates from the server are stalling.
