Cloudflared hosted in Docker container -- Sporatic network latency with cloudflared ->local services
This very well may be more of a Docker issue as opposed to Cloudflared, but I haven't witnessed it firsthand with anything other than Cloudflared yet.
For a while now (before the release of 2024.6.1 of cloudflared), I've noticed that if my cloudflared container doesn't obtain the 172.22.0.2 address on the non-default bridge network it's assigned, I'll experience intermittent, significant latency when accessing services via the cloudflare tunnel. When the request succeeds, but is delayed say up to 15-30 seconds or so, there's no indication in the logs that there's any issue. Occasionally, there will be enough latency that the call will timeout return a 530. I do see the occasional "i/o timeout" in the cloudflared logs from cloudflared to the locally running service in the scenario where the connection fails completely. Now, if I stop whatever container acquired the 172.22.0.2 address, stop cloudflared, start cloudflared (so it picks up the 172.22.0.2 address) and then restart the other container that had the address prior, zero issues. Everything is super fast without latency.
Here's a look at a response time graph of a local service via cloudflared where I hadn't realized this issue until 5 am this morning. Once I did the switch described above, smooth sailing 😕
For a while now (before the release of 2024.6.1 of cloudflared), I've noticed that if my cloudflared container doesn't obtain the 172.22.0.2 address on the non-default bridge network it's assigned, I'll experience intermittent, significant latency when accessing services via the cloudflare tunnel. When the request succeeds, but is delayed say up to 15-30 seconds or so, there's no indication in the logs that there's any issue. Occasionally, there will be enough latency that the call will timeout return a 530. I do see the occasional "i/o timeout" in the cloudflared logs from cloudflared to the locally running service in the scenario where the connection fails completely. Now, if I stop whatever container acquired the 172.22.0.2 address, stop cloudflared, start cloudflared (so it picks up the 172.22.0.2 address) and then restart the other container that had the address prior, zero issues. Everything is super fast without latency.
Here's a look at a response time graph of a local service via cloudflared where I hadn't realized this issue until 5 am this morning. Once I did the switch described above, smooth sailing 😕
0 Replies