When something's broken on the web, the hard part is usually figuring out where. The developer says it's the server, the host says it's the code. I work across all of it, so I find the actual cause instead of guessing.
Same-day start on emergencies. US Eastern time, remote, clients anywhere.
I look at all four
Nobody in that conversation is lying. Each of them can only see one layer. The bug is usually sitting between two of them.
One-hour fixes are welcome. So are the problems that have already beaten two other people.
Recent saves
A few from the fleet I run day to day.
Edge → Server
Long pages stopped partway through on one environment, with a 200 status.
The cause was NGINX workers running as a different user from the one that owned NGINX's temp directories. Any response too big for the memory buffers got cut off when NGINX tried to spool it to disk.
Server → JVM
JVMs across the fleet were freezing.
The cause was a pinned young-generation size in the G1 garbage collector settings. Pinning it takes away G1's ability to size the young generation to meet its pause-time goal.
Code → JVM
The first hit to a page was slow on every server in the fleet.
The cause was 9,933 orphaned object-pool MBeans left registered in JMX, leaking Metaspace. Tracked down and resolved fleet-wide.
Rates
Site down, pages crawling, errors nobody can reproduce, a deploy that went sideways. Call or text and I start the same day. Most of these are found inside two hours.
Bugs in Node, Lucee or PHP code, slow queries, upgrades off end-of-life versions, and the backlog of small things nobody has gotten around to.
Porting a CFML application to PHP (Laravel or CodeIgniter) or Node, or onto a supported Lucee stack if a rewrite isn't worth it. Done in stages so the site keeps running while it moves.
A written assessment of your code, database, and servers: what's out of date, what's exposed, whether your backups actually restore, and what breaks first under load. You get a prioritized list with effort estimates, whether or not I do the work.
You get my number before anything breaks, priority response when it does, plus monitoring, patching, backup checks and a standing block of hours. For teams that need someone to call without hiring for it.
How I work
I'll tell you when your system is fine. That conversation costs me the work, and I have it anyway, because I'd rather be the person you call next year than the one who billed you the most this month.
Changes come from version control, not an undocumented shell session. You get the code, the scripts, and documentation for everything I do, so the next person can run it without me.
I've been doing this for 25 years. Day to day I run 53 Rocky Linux servers serving more than 600 production sites, including the custom OpenResty/Lua firewall in front of them. I recently took a B2B marketplace from a legacy CFML codebase all the way to Node in production, and I write PHP in Laravel and CodeIgniter, so moving off ColdFusion is familiar ground. BS in Computer Science, MBA with an IT concentration. I write up the interesting failures on the blog.
Contact
If something is down right now, call or text. Otherwise send the details and I'll get back to you the same day with what I'd do.
Direct