Yes, we experienced a downtime too, but we are running also on non-metal servers in EU ... downtime was between 15:44 and 15:50, services were unreachable, "Application crashed"-page from Railway was shown.

Same problem here, at around 15:50 our services were not reachable. Were using EU metal-servers, switched them back to GCP for now.

Hello,

We have investigated and identified the cause, related to our redundant interconnect between Railway Metal and Google Cloud Platform in EU West.

From 14:45 UTC to 14:50 UTC, we observed a drop in traffic on this interconnect:

We are following up with GCP and the interconnect provider to determine what happened. Please note that this issue is not related to Railway Metal itself. Most traffic destined for Railway metal is still forwarded via our HTTP proxies running on Google Cloud, which is why Railway Metal users were effected by this incident.

We are currently in the process of rolling out our full-metal anycast edge network, which will mean in the future we won't have to rely on GCP or this interconnect at all. It's in beta, but if you'd like to move your domains to our new edge network, you can click the "bolt" icon located on your domain list.

I would like to stay away from these Beta's. If we have all our services on non-metal or all on metal does that mean no interconnect is used?

Service on metal, domain on GCP edge = Interconnect is used

Service on metal, domain on metal edge = No interconnect, full Railway Metal

Service on GCP, domain on metal edge = Interconnect is used

Service on GCP, domain on GCP edge = No interconnect, full GCP

Allright, thanks for the info