5 months ago
I'm not seeing any discussions related to this in the Discord..
Around 1-2 hours ago, I was notified of 2 things:
-
Hostname/IP does not match certificate's altnames: Host: unproxied-custom-domain-here. is not in the cert's altnames: DNS:*.up.railway.app
- HTTP 520 responses for Cloudflare proxied custom domains
Looking further, there was also a spike in latency
This affected all of my services, across multiple projects, and all of the deployments were healthy w/o redeploys
The status checker I'm using here is hosted in AMS
Incident started around 01:15 UTC and ended around 01:25 UTC
Attachments
40 Replies
5 months ago
What do your http logs say?
5 months ago
Were all green, let me check again
5 months ago
Scrolling through the HTTP logs on a service that got the certificate issue, it's all 200s
5 months ago
Link
5 months ago
noticeable gap (utc+2 tzs)
Attachments
5 months ago
linked, should be this one
5 months ago
To clearly clarify, this is no longer an issue
5 months ago
Attachments
5 months ago
I'm not seeing anything on our end here.
5 months ago
in case missed
5 months ago
Yes, I've looked at our monitoring for the time you gave.
5 months ago
another example, cloudflare proxied
Attachments
5 months ago
Not seeing anything on our end for that either.
5 months ago
I mean, could it be an observability issue? I don't think Cloudflare lies about 520s
5 months ago
I am honestly not sure.
5 months ago
Not a new issue either I believe
Attachments
5 months ago
I'm also not seeing other users report this 🙁
5 months ago
"maybe I'm just not like other users"
5 months ago
think about the times where I called out an outage first (at least in the discord)!!
5 months ago
Haha I'm not sure what you want me to say to that.
5 months ago
didn't expect you to
5 months ago
I'm fine with leaving it at this, no SLA or critical services on my end
5 months ago
may be worth looking into (deeply) though..
5 months ago
I've looked, I don't see anything from any of our monitors.
5 months ago
what about now!
5 months ago
I'm also observing this (second time today a couple minutes ago) - first time was 8:20 UTC
Attachments
5 months ago
^
I've been seeing similar behaviour just now.
UTC timeline of the recent flapping I observed (triggered by my monitoring system; which is also monitoring services outside of Railway as well and only Railway was impacted:
08:21:35 — DOWN
08:22:56 — UP
08:24:58 — DOWN
08:26:18 — UP
09:40:46 — DOWN
09:44:44 — UP
09:58:27 — DOWN
09:59:47 — UP
10:01:53 — DOWN
During this window I did some diagnostics:
DNS resolution was normal and stable (single IPv4 A-record; no IPv6).
TCP connect to port 443 succeeded immediately.
The failure happened during TLS negotiation: curl sent the TLS ClientHello, then timed out waiting for the server’s ServerHello (connection established, but the TLS handshake response didn’t come back).
Shortly after, the alert cleared by itself. I re-ran the same diagnostics and TLS completed normally (TLS 1.3), ALPN negotiated HTTP/2, and the health request returned 200.
Note I am NOT behind CF proxy.
5 months ago
"Same" and "this" meaning you all have seen a cert for a service domain instead of a custom domain?
5 months ago
Hello!
We've escalated your issue to our engineering team.
We aim to provide an update within 1 business day.
Please reply to this thread if you have any questions!
Status changed to Awaiting User Response Railway • 5 months ago
5 months ago
Happy? 😆
Attachments
5 months ago
I win
5 months ago
<:proud:800594749714071553>
5 months ago
Hello,
We have provisioned more capacity in regards to our edge network in the EU-West region. You should no longer see these blips of incorrect certificates.
5 months ago
Thank you.. how can this be caught automatically next time?
Status changed to Awaiting Railway Response Railway • 5 months ago
5 months ago
We are going to alert on the graph I shared above, and long term, we are well underway on a complete rewrite of the network control plane that will outright solve for this, and solve for many other potential issues.
Status changed to Awaiting User Response Railway • 5 months ago
Status changed to Solved brody • 5 months ago