Enjoy haha

Is the source repo for @cloudflare/worker-sentry lurking anywhere?

The package would be super useful if it didn't rely on having to use wrangler vars 😛

its short enough you could just copy paste https://cdn.jsdelivr.net/npm/@cloudflare/worker-sentry@1.0.0/index.js

Don't know about that particular package, but there is this: https://github.com/robertcepa/toucan-js

looks like its a wrapper of that

@cloudflare/worker-sentry doesn't seem to be listed anywhere on here https://cloudflare.github.io/

Sneaky haha

Yah, @cloudflare/worker-sentry is a wrapper around it

I've just yoinked the source and modified, but would be good to update the original to take the env stuff in as an obj

IT WORKED

hype

Hello!

Always wanted to ask if Cloudflare ever had plans of moving into email type business. Would love some email forwarding with Cloudflare registrar

Interested to see if Cloudflare will open the ports for workers to be able to handle emails

hi

hi

Anyone know what these spikes are when all the requests require basically the same workload? Can somebody shed light on how these CPU time are calculated, and why might requests coming from one country take significantly longer in CPU time than requests coming from another country? https://imgur.com/a/pjXFWZe

If some users in a specific geographic region have higher latency to CF and slower connnection speeds (meaning they are connected longer), will that require more CPU time? Or is the whole request returned by the worker and handed off to a buffer until the users connection can finish transferring the payload (meaning no additional CPU time regardless of their latency of connection speed)?

Those are most likely cold starts.

And what about why requests from on country (specifically China) have a significantly higher CPU time than requests from other countries (Singapore, USA, etc...)

Higher latency is that the requests from Chine go outside of China, higher CPU I have no clue. Probably less average bandwith (with higher latency) hence more time to process the whole thing?

Hence why im asking if somebody can shed some light on how the CPU time is measured. For example with measuring CPU time in PHP it only counts the time executing your script, none of the time spent by apache up until the point of your script execution time is counted, and no time spent by apache after the buffer the closed is counted. So i’m wondering specifically how much of the chain is counted as CPU time in Cloudflare Workers

That is a good question I would also love to know answer to 😄 However never ran into issues even with occasional >10ms or >50ms (depends on the plan)

I’m not running in an enterprise plan so my users in China (me included) will be getting served by a node outside. And considering these are the biggest user group who seems to have a large discrepancy in CPU time compared to the rest if the world users then I assume latency & connection speed have a factor in the way workers CPU time is calculated.

The difference can be from half a millisecond with USA users to being 10 to 20 milliseconds for China users.

Fluctuations on day and time. I sometimes see my speeds to cloudflare go from gigabit/sec to less than a megabit/sec

I haven’t seen anything in the 50ms as you say. Thats pretty huge

Afaik even with Enterprise plan you would need to opt-in for Cloudflare in China network, as there is the need of ICP license first, in order to be served from China mainland. I guess it will be connected to the fact the requests are just taking longer (as the nearest cloudflare pop is outside of china) and the worker waits - not an expert, just rough guess. But I would love to see/learn more details on it 👍

I’m aware. I run a dev business in China. Hopefully someone at Cloudflare will write a blog about where along the chain the CPU time calculation starts and stops.