Repository navigation
[Due for payment 2026-08-28] [No QA] stop the pong watchdog from reconnecting Pusher #98886
Description
Problem
The PING/PONG watchdog declares a live socket dead. Its 60 second staleness threshold equals the PING cadence the app actually reaches in production, so the check fires on healthy connections. Every false alarm discards a working socket and pulls a full ReconnectApp sync behind it.
The volume is large. In a 4 hour baseline window the presumed dead path fired 96,667 times across 20,163 users, an average of 4.79 each, with 541 users firing 20 or more times. That works out to 565,054 watchdog reconnects per day and roughly 1,450,000 presumed alive log lines per 4 hours.
Solution
Delete the reconnect from the watchdog. Recovery moves to the Pusher SDK, which detects a dead socket with a websocket ping frame on the connection under test, so a stale PONG no longer throws away a socket that still works. The watchdog keeps logging staleness, with [Pusher PINGPONG] The server has not sent a PONG in ... leaving recovery to the Pusher SDK replacing the old presumed dead and Pusher is being reconnected line.
The prediction this PR is measured against: presumed dead fires, watchdog reconnects and presumed alive log lines all drop to zero, and pushed ReconnectApp per user per 4 hours falls from 5.45 to between 4.25 and 4.47, a drop of 18% to 22%. GetMissingOnyxMessages did not advance is the control at 1.10 per user and must not rise. Recovery is slower by design, so allow up to 150 seconds for a socket to come back after sleep or backgrounding.
Activity
[-][No QA] stop the pong watchdog from reconnecting Pusher[/-][+][Due for payment 2026-08-28] [No QA] stop the pong watchdog from reconnecting Pusher[/+]melvin-bot commentedon Aug 21, 2026
@ZhenjaHorbach
The solution for this issue has been 🚀 deployed to production 🚀 in version 9.4.57-3 and is now subject to a 7-day regression period 📆. Here is the list of pull requests that resolve this issue:
If no regressions arise, payment will be issued on 2026-08-28. 🎊
The following checklist (instructions) will need to be completed before the issue can be closed. Please copy/paste the Contributor+ Checklist from here into a new comment on this GH and complete it. If you have the K2 extension, you can simply click: [this button]. If no checklist is needed for this issue, you can click: [no checklist button]
melvin-bot commentedon Aug 24, 2026
@ZhenjaHorbach Uh oh! This issue is overdue by 2 days. Don't forget to update your issues!
melvin-bot commentedon Aug 26, 2026
@ZhenjaHorbach Eep! 4 days overdue now. Issues have feelings too...
melvin-bot commentedon Aug 27, 2026
Triggered auto assignment to @mallenexpensify (
Awaiting Payment)melvin-bot commentedon Aug 27, 2026
Payment Summary
Resolving PRs:
[No QA] stop the pong watchdog from reconnecting Pusher #98884
Contributor: @adhorodyski does not require payment (Contractor)
Reviewer: @ZhenjaHorbach owed $250 via NewDot
BugZero Checklist (@mallenexpensify)
ZhenjaHorbach commentedon Aug 28, 2026
It's a performance issue
So don't need a checklist here
mallenexpensify commentedon Aug 31, 2026
Payment Summary
Contributor+: @ZhenjaHorbach due $250 via NewDot
garrettmknight commentedon Sep 2, 2026
$250 approved for @ZhenjaHorbach