I will fix socketio scaling, disconnect and memory leak problems


About this gig
Your WebSocket layer was fine at 50 users and started dropping connections, duplicating events or leaking memory at 5,000. That is almost never a bug in Socket.IO itself. It is missing sticky sessions, no Redis adapter across instances, unbounded room state, or reconnection logic that re-subscribes without cleaning up after itself.
I have spent four years building and running real-time platforms where thousands of users hit the same servers at once, as the only engineer accountable at 2am when it broke.
WHAT YOU GET
Basic - Diagnosis. I review your Socket.IO, Redis and load balancer setup and send a written report: what is breaking, why, and what each fix costs in effort. 2 days.
Standard - Diagnosis plus one root cause fixed and tested against your load. 4 days.
Premium - Full remediation: Redis adapter, sticky sessions, room and listener cleanup, reconnection and state resync, with a before-and-after load test. 7 days.
HOW IT WORKS
1. Send repo access or your socket and server files, plus the concurrency where it starts failing.
2. I reproduce it, or tell you within a day that I cannot and cancel rather than guess.
3. You get a pull request with the reasoning written out.
Get to know Harsh
Real time backend engineer, Socket IO, Nodejs and Redis at scale
- FromIndia
- Member sinceOct 2020
Languages
Hindi, English
Other Software Development Services I Offer
FAQ
Can you work with our existing code even though it is messy?
Yes. Real-time bugs almost always live in code that grew faster than it was designed.
What if you cannot reproduce the problem?
I tell you within 24 hours and cancel the order rather than bill you for a guess.
We are not sure whether it is the backend or the client.
That is a normal starting point and the Basic package exists to answer exactly that question.
