And as soon as I post this…. back.

Although, still very sporadic 😕

Yeah my website down too

My app is down too but only in the European regions

experiencing the same thing too

we still experiencing this? its only happening on our railway-edge hosted services, frequent timeouts

cc: @blank (revived)

We were able to get it working with a workaround:

• Ensure runtime is set to legacy

• Deploy on a custom domain

• Now have access to legacy proxy > Enable Legacy Proxy

This caused major disruptions in our production services

Hi all, yesterday we saw a large increase in traffic originating from europe causing high latency only in that region. We did not create an incident at the time because we took steps to mitigate and saw the system recover. Once we got back from the weekend we saw that the traffic was degraded though but the alarms didn't page again.

Here is the incident created https://status.railway.app/cm0b88ku6003wbs533smpr831

if anyone is still having issues with their traffic let me know. There shouldn't be any impact for the past couple of hours.

@Mig having issues

Please see <#1277840390233526273>

issues mentioned were not timeout related^

anyone seeing this please only let me know about timeouts. However, this specific incident is 100% resolved at this time so you should just start another thread.

@Mig I have been getting websocket timeout issues across my projects a fair bit too, but because they are serverless they tend to go away, but sometimes I get a serverless function crashed issue from vercel on my production frontend

This has been occurring for at least a month, so I wasn't sure if it was perhaps a Surreal issue or maybe an issue with Railway

the edge network will close idle connections if no data is sent for some time.

our legacy edge network I believe limited connection also to 10 minutes but don't quote me.

these wouldn't cause timeouts though but that's my only guess for websocket communication issues since they are long kept.

I believe the Surreal database sends pings so that should prevent that? @Mig

our legacy edge network I believe limited connection also to 10 minutes but don't quote me.

If that is true, that sounds about right - I think we were having about 10-13 minute intervals before they crashed (but I can't remember if this was local surreal or production database)

the pings would keep the connection active

I'd have to dig into the config/source to confirm the 10 minute timeout and cannot do that right now.

hey @Albert friendly reminder not to ping team members or moderators.

<#727685388893945877> #5

apologies