Elevated number of 524
Hey guys, we are seeing a lot of 524 responses, while the origin looks completely healthy. Any clues?
CWith Workers? Is your origin taking longer then 100s to respond (long processing job, etc)?
DNot really, our origin is Backblaze
DBackblaze doesn't see any observable issues
DHere is example ray id 86ec4785b2ce2dec
CThis is #workers-help, you're using workers to fetch Backblaze or where exactly do they come in?
DYes, request goes through CF worker
DAnd we get 524 too fast
Dabout 10 s
Cah ok, does your worker do anything else other then fetch backblaze?
There is currently an issue open with Image Resizing if you are using that, resulting in 524s: https://www.cloudflarestatus.com/incidents/p5xk648mwb66. Other then that there's no current issues, would try requests to Backblaze without CF in the middle/try to narrow down the issue
And we get 524 too fastCloudflare pools connections and backs off when your origin is slow/unresponsive (https://blog.cloudflare.com/how-pingora-keeps-count), doesn't have to be after 100 seconds.
There is currently an issue open with Image Resizing if you are using that, resulting in 524s: https://www.cloudflarestatus.com/incidents/p5xk648mwb66. Other then that there's no current issues, would try requests to Backblaze without CF in the middle/try to narrow down the issue
DAfter 4h of back and forth with support
DAs far as I can tell issue started around 9h ago
DAnd yes - worker fetches image and scales it
Cah ok ouch, at least there's a status post now
Here is example ray id 86ec4785b2ce2decbtw green names like me aren't employees. We can escalate issues if they are impactful and replicable though. I can't do anything with a Ray ID, and even to CF Ray IDs are sometimes not helpful/they don't log everything for each request
DYep, I mean - support ticket in CF account. It just took a while to push the fact that there is the issue through
DI really appreciate your help here
DEven trying to listen and understand the problem
