Alin Dima

Tested in versi, here are the performance results: https://github.com/paritytech/polkadot-sdk/issues/598#issuecomment-1792007099

> Discovered an issue that needs fixing: pov-recovery in cumulus does not work if the collator uses an RPC relay chain node. It fails because the partial relay chain node...

opened RFC for discussing the details of the chunk index assignment function: https://github.com/polkadot-fellows/RFCs/pull/47

I had to do a little force push because I messed up a merge commit. That's also why the CI bot requested reviews from a bunch of groups that were...

@ordian @alexggh @sandreim This is ready for another review, I implemented the updated proposal from the RFC. I only have to implement some zombienet tests and adapt subsystem-bench for it,...

Some results from subsystem-bench: with regular chunk recovery: `CPU usage per block 39.82s` with recovery from backers: `CPU usage per block 16.03s` with systematic recovery: `CPU usage per block 19.07s`

I just thought of another improvement to this PR (which I'd like to do as a follow-up, because this PR is already huge). We can cache the req-response protocol version...

I've conducted the final versi testing before merging this and it everything looks good! When enabled, total recovery time is halved and the cpu consumption associated to erasure coding is...

> I don't really think we overload the backers, but we do try backers as first option, and if that fails we try systematic chunks. Going back to backers doesn't...