We had some very degraded performance for about 2 hours today on only one of our apps. Were there any known disruptions that could have caused this?
Experiencing something similar, very slow loads. I think is DNS related. I’ve seen some 502 responses on the statics assets we deploy to Fly.io with [static] option.
Hi folks, my theory is that your apps might be doing some CPU-intensive processes which result in throttling (CPU Performance · Fly Docs). You can check this yourselves in your app metrics page, the CPU page has a section on CPU load and throttling which you can check to see whether it matches the periods where you saw slowness.
I can’t check in more detail without machine IDs, if you care to share them I’m happy to have a second look!
@Moreno what makes you think it’s DNS-related?
- Daniel
It was only happening to certain assets some were loading fine and some other not (the assets that were deployed via [[static]] on fly.toml), cpu was very low and not throttled.
That’s what made me think it was caused by some DNS issue with the static provider you got or something. It was just like 15-20 minutes of issues.
I think in our case, your suspicion was correct - we were being throttled. Thanks for pointing me in that direction.
Actually, the [[statics]] are served from within your Machine itself—which is pretty non-intuitive. The following clarification was added to the docs relatively recently:
If the requested file exists in your
guest_path, the request is routed to a static file server running inside your machine. This bypasses your main web server process.
(boldface added)
Unless you were using the tigris_bucket option, there really isn’t an external provider involved.
Perhaps it was a Disk I/O throttle instead? That limit is as low as 8 MiB/s when serving from the ephemeral filesystem (i.e., the image itself), and I don’t think it’s reported as part of the CPU throttling series, 
…
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.