Simon Cooper

Are there any situations where `_all` *should* change settings on system indices?

I've been unable to reproduce this on 8.3.2 or latest 8.5.0, the settings command completes successfully with the data & transforms in place.

I added some sample data from kibana to a running 7.17.6, added a random transform, then copied the data directory into a 8.5 snapshot directory, started it up, then ran...

We should also update `docs/reference/modules/http.asciidoc`

I am currently reviewing the impact of these changes on performance, GC, and memory usage to lucene & elasticsearch. If this does show an improvement, I'll set this PR to...

For reference, this is aimed at #11711 and #11712

This is ready to review. I've removed the threadlocal byte buffer. With testing in elasticsearch, this reduced the duplicated strings in these fields by 90%. For the linked tickets this...

Another option is to have the `FieldInfosFormat` classes handle string pooling themselves, leaving `DataInput` unchanged.

Unfortunately that doesn't seem to have much of an impact, from what I can see here. @rmuir Would you be against having a string cache specifically in the relevant methods...

To be clear, are you referring to the extra memory used by the deduplication hashmap for the duration of the deserialisation, that will then be eligible for GC after the...