🥴

haha good point, i believe it's referring to lambda tho. in lambda splitting out your 100MB lambda into 5x20MB lambdas can be a very beneficial thing for cold starts. not particularly relevant to workers

doing the lords work

for each worker, I end up with at least 2 versions of it, sometimes three. (dev, staging, and production)

sub 15ms

is that similar to standard worker cold starts?

we currently appear to have some workers publish lag

soooo i will tell you when my new version is actually live

much faster

a normal

fetch(workerurl)

is like 30-60ms

(this is repeated invocations so it'll be warm)

bindings = nyoom

Is there a way to see how long it took my Worker on average to return a response?

not how long I got billed

but what the user felt

cause health analytics is showing my Worker takes a whooping ~150ms on average

Maybe with RUM?

But that necessitates client-side JS to measure.

I want to see like the RTT if I used curl too

That would require network speed metrics, which you basically can’t get without a client-side script(or running a ping toward the users IP address)

At most you can know when a Request enters the CF edge, and when the Response leaves

that's what I want to know

Might be able to do that with LogPush. Do you have Enterprise?

Pro

Don’t think you can do it then

hmm

I see EventTimestampMs in logpush for the start of the Worker I think

but doesn't give me the end time sadly

Yeah, I think you need full LogPush to measure that, which is Ent and above, afaik