server was unreachable until re-deploy

hey Fly friends. Looks like sometime overnight nikola-sharder went down without me having done anything. Updown.io reported it as unreachable. I’ve re-deployed now and it appears to be coming back up. Any ideas on what happened and how I can prevent this in the future? Thanks!

perhaps related to the email I just got about " Routing issues with 50.31.246.0/24 IP block"

Yes that was it. This is currently resolved, and we’re monitoring the situation. Also digging into what happened… will update the status page soon!

We haven’t implemented this yet, but that IP block is going to get all our free customer traffic soon. You should get IP in a different block if you can!

You can run fly ips allocate-v4 you’ll get another. If your DNS is using a CNAME to your <app>.fly.dev app, you don’t need to make any changes. If you’re using an A record, you’ll need to update that. Then run fly ips release <oldip>.

1 Like

thanks! using CNAME to .fly.dev, so that means a complete no-op?

Yep! Just give it a few minutes after you add the new IP before you remove the old one.

ah got it, so still need to run fly ips allocate-v4

silly q: what are the benefits of being on a different IP from free traffic? Like I have some guesses, but I’m not sure.

My guess will be that the free traffic ip block will start reusing IP addresses as I’m pretty sure they hinted at it before.

Yeah a little of that! We reuse IPv4 addresses when people delete apps and free apps churn pretty frequently. You never know how someone dirtied an IP.

Free services attract people who are prone to getting DDoSed. We mitigate DDoS attacks a whole /24 block at a time. Today’s issue was caused by a DDoS mitigation bug. You won’t notice most DDoS attacks, but it’s still worth avoiding higher risk IP space.

1 Like

Thanks Kurt!