Hi there!

We're experiencing persistent DNS resolution failures when Inngest tries to reach our worker functions hosted on Railway. This is severely impacting our production batch jobs with a 3-5% daily failure rate. The failures happen before our worker code even executes, pointing to infrastructure-level connectivity issues.

I have contacted Inngest, and they told me it is a Railway issue where Railway is throttling the Inngest DNS requests. They said it's RW's job basically to remove the rate limit and urged me to reach out to you instead.

We are actively loosing customers over this, so I'd really appreciate your help before we need to migrate away.

Error Pattern:

Error performing request to SDK URL: error performing request: 
lookup <worker-hostname> on [2607:e3c0:a040:9002::64:ff9b]:53: **server misbehaving**

dial tcp [64:ff9b::4221:16ef]:443: i/o timeout
Error performing request to SDK URL: error performing request: lookup <worker-hostname> on [2607:e3c0:a040:9002::64:ff9b]:53: server misbehaving dial tcp [64:ff9b::4221:16ef]:443: i/o timeout

Key Technical Details:

The 64:ff9b:: prefix indicates NAT64 (IPv6-to-IPv4 translation) but I'm honestly not sure. Railway logs show zero incoming requests at failure timestamps - DNS fails before HTTP connection Our Railway service is healthy (low CPU/memory, responds fine to direct requests) Failures are sporadic - successful runs happen immediately before/after failures Affects all our scheduled and event-triggered functions

Impact: This is a blocker for scaling our usage. We need reliable execution for production workloads, and the current 3-5% failure rate makes it difficult to commit to a paid tier. The intermittent nature suggests DNS resolver or NAT64 gateway instability on Inngest's side (at least after hours of AI debugging with Opus 4.5 we came to this conclusion) when connecting to Railway.