Uptime Monitoring alerts from Uptime Kuma shows timeout of 48000ms exceeded (only with Cloudflare enabled?)
What steps have you taken to resolve the issue?
Here are the hard facts:
I have Uptime Kuma set up on many websites - usually I do checks every 2 minutes, for troubleshooting I might go down to 1 minute
Out of a sudden random hourly downtimes show up according to Uptime Kuma
Those downtimes are hardly ever noticeable in real life (i.e. visiting website from browser)
Those alerts can show hourly, every few hours, it’s quite random
It’s not new, I have seen this issue in the last year several times, never really solved it, sometimes it just stop by itself sending me those alerts
Hosting provider support is always saying it’s everything else except for the server (they might be right sometimes)
In general I see this again and again with different websites. Recently it showed up again with a website (or I rather should say with an entire domain).
→ I never know if it is an actual issue or a false positive downtime
→ I never know if the WordPress website internally has problems, or it’s an issue with hosting, Cloudflare or something in between
In this particular case I tested a lot. I even migrated to a different server (copy of the website) to see if it makes a difference, but it does not. So it’s pretty clear that it is not the hosting provider.
When I pause Cloudflare, it seems the alerts from Uptime Kuma are gone. So what related to Cloudflare leads to “[Down] timeout of 48000ms exceeded”
Of course I have also many other websites active where it does not show and Uptime Kuma works as it should.
Thanks - this problem goes on for many years, so any input can be very valuable.
I do not have bot protection enabled, because I cannot modify it at all. Bot Fight Mode is either on or off. If it’s on, Uptime Kuma will be also blocked, therefore I don’t use it at all.
I studied the logs quite a bit (honestly not easy) to figure things out. I also allowlisted the IPv4 and IPv6 of Uptime Kuma on the particular site where the problem showed up again, so it should not be blocked and if it was blocked it would impact the uptime monitoring on a global level for all my WordPress websites and Cloudflare accounts.
I am also not alone in this. Many people facing the same problem over many years. On the one hand I can prove that as soon as I pause Cloudflare the issue is gone. On the other hand two users reported they are facing the same issue without Cloudflare.
That means, Cloudflare itself might not be the main source but it’ll add a lot to it and can be the main trigger with this problem.
I summarized a lot on this entire issue here earlier today. My theory would be, if Cloudflare is in the game then this adds so much latency that it’ll most of the time be the main trigger to exceed the 48 seconds and therefore run into a timeout. There might be other sources as well (e.g. docker / DNS lookup) but it seems Cloudflare is the main driver of the problem here.