Nick Mills-Barrett

One alternative I'm considering using - POST the script from git during deployment using the webUI's API (https://github.com/binux/pyspider/blob/master/pyspider/webui/debug.py#L169). Won't protect against further changes via the UI but updates w/git changes.

~~Ok this is now confirmed working on all platforms, so implementation is done?~~ See description!

Based upon internal discussion we no longer think this is the right approach and thus we are closing this MSC in favour of [MSC3860](https://github.com/matrix-org/matrix-spec-proposals/pull/3860) & [MSC3870](https://github.com/matrix-org/matrix-spec-proposals/pull/3870). I will leave the...

I don't think we're missing any indices, the query plan seems sensible to me - and during usual running it's not a problem. We have one particular large room that...

The other optimisation that comes to mind, particularly for rooms like this, is process by room not by user but that’s a huge change in how things work and probably...

Been poking around this some more, one potential issue is [read receipts don't always have a matching event in the DB](https://github.com/matrix-org/synapse/blob/06df5d4250f54d5a95b0c90bfc9352ec6f02c520/synapse/storage/databases/main/receipts.py#L641), I suspect this is possible over federation that the...

> That query plan looks like its pulling out all receipts for a user. We should be able to refactor it to only check receipts for rooms that have push...

> Been poking around this some more, one potential issue is [read receipts don't always have a matching event in the DB](https://github.com/matrix-org/synapse/blob/06df5d4250f54d5a95b0c90bfc9352ec6f02c520/synapse/storage/databases/main/receipts.py#L641), I suspect this is possible over federation that...

Progress update: we are using https://github.com/matrix-org/synapse/pull/13918 in our fork and this has yielded _significant_ improvement to the IOPs at all times and in particular during posts in large rooms. Next...

Out of interest - how many federation senders are running on matrix.org? I wonder if something is weird about our setup; we have 3x of them each doing ~1k/s on...