chore: delivery monitor for store v3 reliability protocol #2977
Conversation
Better structure of delivery monitor and related modules store protocol: simple cleanup of unused const use of observer observable pattern to inform delivery_monitor about subscription state waku_store resume does not need to get an explicit peerId as query param waku_sync protocol needs some peerId as peerId is optional in query proc send_monitor becomes a publish observer of lightpush and relay deliver monitor add more protection against possible crash and better logs
|
You can find the image built from this PR at Built from 381b15c |
Ivansete-status
requested review from
SionoiS,
gabrielmer,
richard-ramos,
kaichaosun and
NagyZoltanPeter
waku/node/delivery_monitor/not_delivered_storage/not_delivered_storage.nim
Show resolved
Hide resolved
Co-authored-by: gabrielmer <[email protected]>
There was a problem hiding this comment.
LGTM Nice work!
If only Sync was better implemented it could have been useful for this feature. Without proper storage though it's difficult to use.
As for re-sending messages can it still work with RLN? If verifying that a msg is missing take too much time the epoch would change and the proof would end up invalid no? That would also happen sometimes at the epoch boundary even if verifying is fast.
ah yes, good point! This is something we will need to fine-tune in the future when RLN is applied |
Description
This PR reinforces the efforts in the proper messages delivery, considering both sending and receiving.
Changes that try to replicate similar changes that are currently implemented in
status-go/go-waku.Notice that I haven't tested it yet. The idea is to start testing, and adjusting its implementation, once we integrate
nwakuinstatus-go.Changes
reliability. Disabled by default.Issue
closes #2819