[mitchellst]

You have encountered a problem too big to solve alone, so your solution involves gaining the right allies and persuading them to do the right things. So, however frustrating this is, remember first always to be kind, to your team and to your management.

To make allies of senior management, you need metrics. You need to show, concretely, how current operations put revenue at risk and make the incremental investment necessary for their roadmap items prohibitive. If you can swing a penetration test, they'll probably find plenty on a stack like this. Then you have a security justification. If not, get the best monitoring stack you can. Demonstrate reliability and performance issues. (As well as reliability and performance improvements.)

From there... I'll say the #1 tool I've used in situations like this is Fastly. VCL is way more flexible than your 10k line nginx rewrite file (I've been there, too). And the edge caching will paper over minor outages. Rollbacks are easy. Rebuild your site piece by piece and stitch it all together with a reverse proxy at the edge.

Advice: propose a "canary" portion of the site to rebuild, and make it the lowest revenue / highest complexity thing you can. Once you stabilize the cash cow, getting the buy-in to finish the job and deprecate the old code base will be tough.

I'd also advocate for adding 1 incremental engineer to your team. Make it a senior dev and interview specifically for people who have done this sort of thing before. Your team needs a hands-on mentor in the trenches with them.

Best of luck. It isn't easy, but it's rewarding.