What does “too many connections” mean?

You fire off a big send and suddenly your logs are full of 421 errors with a message like "too many connections" or "connection refused." It's not a reputation problem. It's a plumbing problem.

When your sending server opens an SMTP connection to a receiving server, it's essentially knocking on a door and waiting to be let in. Receiving servers only keep so many doors open at once. If your infrastructure tries to open 50 simultaneous connections to the same server, and that server's limit is 10, the extra 40 get slammed shut before a single message goes through.

Receiving servers set these limits for a few practical reasons. They're protecting their own resources (every open connection costs CPU and memory). They're keeping things fair so no single sender monopolizes capacity. And they're limiting the blast radius if someone starts sending spam at high speed.

The fix isn't to retry harder. It's to slow down and queue smarter.

• Reduce your max concurrent connections to that domain. Most sending platforms let you set a per-domain concurrency limit. A safe ceiling for most receiving servers is 5 to 10 simultaneous connections.
• Stagger your sends. Instead of blasting all connections at once, send in smaller batches with short gaps between them.
• Check your ESP's connection settings. Platforms like Postmark, Twilio SendGrid, and Mailgun expose concurrency controls. If you're using a managed ESP, they may handle this automatically, but it's worth confirming.
• Don't confuse this with rate limiting. Rate limiting caps how many messages per minute you can send. Connection limits cap how many sessions can be open at the same time. You can hit both independently.

If these errors are popping up regularly, it usually means your sending infrastructure is misconfigured for the volume you're running. It's a fixable issue, but it does require getting into your sending settings. If you're not sure where to look, our SOS hotline is free and we're happy to walk through it with you.