Basically, we use envoy as our proxy and our team needs to support a new protocol that envoy doesn't support. I voted to write a custom plugin/extension to envoy to support this new protocol. But the rest of the team wanted to "write" a brand new proxy, from scratch, because they think it would be easier to debug issues since all of the code is written in-house. Discussions were getting heated, and in the end, we ended with a vote. Obviously I lost because everyone else wanted to write a brand new proxy. One month in and we're not getting anywhere. We're basically writing a new proxy with knowledge of tip of the iceberg. We have to go into beta after Thursday. We have a few happy paths working, but a long way until we're comfortable with this code in production What would you have preferred? Writing a plugin or rewriting a whole proxy? Tc 160
Classic 😂
If I were you, push for stronger momentum but interview asap
I tried, but we ended with a vote because the discussions weren't going anymore. And job market is so bad that I dono if jumping ship is even possible
Aren’t there more senior technical leaders you could have brought into the discussion to help influence? This is a pretty big decision and risk to take. This is where you need to build that structure or network so you can have more people weigh in. At least then the risk is more widely understood if nothing else. I mean too late now it sounds like, but for the future.
This is why we’re fucked. The engineering leaders and directors push for shit like this. They rewrite shit just to rewrite shit. They’ll spend a year or more “staying busy” with this and later they’ll cheer they delivered a another sandwich. This needs to end and the engineering leaders that do this fired. There is one in particular that needs to go for not delivering what he’s supposed to and that’s JR. He’ll do anything but what he’s supposed to including take ideas from everyone and then let the whole team work on it as a learning experience while the main project burns. Also they insist everything be written bespoke in house. It’s nonsense.
A full fledged web proxy? What are we - Cloudflare? 😁 https://blog.cloudflare.com/how-we-built-pingora-the-proxy-that-connects-cloudflare-to-the-internet Really depends on the use case. But without details I’d say - update your resume - the people who chose to write a proxy from scratch did it to update theirs. And get out before it’s a failure.
My org created a PaaS that does exactly what k8 does a few years ago and it's still running
But did they do it before k8s was widely adopted? Then maybe… Or did they do it after, but do a better job than k8s in your opinion?
Rewrite may not be the issue. But the estimate is off like shit. A techlead should have called out the risk
This is what happens when you have too much democracy in team. Where was your architect in this discussion?
Knowing some of the eng teams we have the PE Archtect was voting for the rewrite for funzies.
The architect voted to rewrite. He's an old timer that doesn't know cloud or open source solutions. And everyone else is essentially sucking up to the architect
Out of curiosity what protocol do you need that envoy doesn’t support?
That's not even a rewrite of your own code. That's a rewrite of an established product. It's like saying hey let's rewrite the OS cause I don't like this button.
Just make sure you are not the scapegoat.
What’s this for?
Proxy load balancer for a web application that cannot be on AWS or other cloud providers
Seems insane to write a whole new thing for this