i would also love to get access to browser rendering beta: `3e38dc7049f501248503b981bf442450`
i would also love to get access to browser rendering beta:
3e38dc7049f501248503b981bf442450
Z
EI just took a look at adding instrumentation to the 
@cloudflare/puppeteer package, I realised that while adding the spans is going to be very straightforward, it would be quite a bunch of work figuring out what to instrument. So if someone else wants to do that, I would be more than happy to help that. But I am going to focus on the core library in the copious spare time that I have 
Con our ent account, im getting
getting the error when trying to run
workers.api.error.no_access_to_browser_worker, do we need to enable it?getting the error when trying to run
dev or deploy. Role is worker admin. Can dm the acc id
7Can develop do something
Cdm'd the PO the be added
ZIt's not enabled by default for any accounts currently. I assume your contact at CF will be able to add it for you, but otherwise you can also put the account ID in here: browser-rendering-apiBrowser Rendering API Beta Access
Cthanks, dont want to post the account id publically, messaged vero
ZAlright. Just FYI: The account ID is considered safe to be public.
Cokta was considered safe 
HI mean, yes, your account ID can be used to in an attack. But, to perform said attack, the attacker needs access to your account. If they have access to your account, they can find your account ID anyway
HIt's like a Discord User ID. Here's mine:
459755459117711362. It can be used in a attack, but it is by no means what determines whether you are vulnerable or notHSame thing with your domain
CIts more around our audit guys, I'm just playing with the okta thing.
I trust cf over any provider
I have browser rendering on my personal account, so can build on that till its enabled on the ent acc
I trust cf over any provider
I have browser rendering on my personal account, so can build on that till its enabled on the ent acc
Ahey, in this code it returns empty json
Abut i just want to get the inner text of selector's
Acan you guys help
AGitHub
Scraping Food List of Istanbul University for goodwill - GitHub - cahmetcan/iuc-yemekhane-scrape: Scraping Food List of Istanbul University for goodwill
WI used this feature to generate OG:Image generation. My server serves html, then worker renders.
TCheck out https://github.com/vercel/satori#documentation
- it's probably faster than spinning up a whole browser to render an image
GitHub
Enlightened library to convert HTML and CSS to SVG - GitHub - vercel/satori: Enlightened library to convert HTML and CSS to SVG
JHello. I'm taking a screenshot of websites with the
screenshot() method, and even with fullPage: false i'm getting the full site instead of just the view port. When I also pass captureBeyondViewport: false i am getting only the viewport, however, I also get the scrollbar - which I don't want. Is there a reason I'm getting the full page even when setting it to false. I'm manaully setting the viewport as well.
AHello people!
AI was told I can request for the beta access over here!
ACan I get access for this Account ID: 6413bfbe023fbf879350a20fe7a2245b?
C
Put your ID in the thread for access here: Browser Rendering API Beta Access (can also be found via pins)
Aalright!
JHello, I'm trying to understand the proposed pricing structure and how to optimize it. Based on the proposed pricing in this blog post[1], you get 500 sessions per month for free and then 10 cents per session after that. A session, as I understand it, is an active running browser. The naive approach, at any sort of scale, quickly becomes expensive. So the blog says to use a durable object to keep the browser alive and re-use the existing session instead of starting a new one. The example code uses an alarm to keep the browser alive while idle for 1 minute before shutting it down. Using this structure, If I'm understanding it right. Worst case scenario, if I had a job that triggered every 61 seconds, I'd end up starting about 1440 sessions per day. Which is $144/day or over $4k/mon. That's expensive. So, the solution would be the extend the life of the durable object to, say 2 minutes, which would reduce the $4k to $0, for the browser cost.
However, the durable objects aren't free, but I'm having a much harder time understanding how expensive they will be. For example, why do I need to shut down the browser at all? How much does it cost for the browser to sit idle in a Durable Object? Is that documented anywhere? And what if it's not idle? If my worker is handling 1 request per sec, so the browser is in active use all of the time, what sort of costs is that going to yield for the DO?
Any information would be appericiated.
[1] https://blog.cloudflare.com/running-serverless-puppeteer-workers-durable-objects/
However, the durable objects aren't free, but I'm having a much harder time understanding how expensive they will be. For example, why do I need to shut down the browser at all? How much does it cost for the browser to sit idle in a Durable Object? Is that documented anywhere? And what if it's not idle? If my worker is handling 1 request per sec, so the browser is in active use all of the time, what sort of costs is that going to yield for the DO?
Any information would be appericiated.
[1] https://blog.cloudflare.com/running-serverless-puppeteer-workers-durable-objects/
MIt costs about $4 to keep a DO alive an entrie month but I'm not sure if that's somehow against terms for browser rendering or if there's unintended consequences.
browser-rendering-api
As far as understanding the pricing I've got no clue and there hasn't been any real clarification on that. Not a single use case makes sense to pay $4000 per month for a single browser instance. You could hire a human for that.
browser-rendering-api
As far as understanding the pricing I've got no clue and there hasn't been any real clarification on that. Not a single use case makes sense to pay $4000 per month for a single browser instance. You could hire a human for that.
JThanks for the reference to those previous messages. I'm glad that I'm not the only one who is confused by all of this.
JSome people were mentioning that having one session open, means one request at a time, but that's not true. You can use
newPage() to use the same browser to process request async - but there's got to be some limit to that. I can't imagine one browser would do 100s of request simulateously. But again, those limits don't seem documented either.
USeeing as the blog post about browser rendering showed how to keep sessions alive with a durable object I would find it highly unlikely to be against TOS 
JThe question is whether keeping the browser open indefinitely is against the TOS. The blog shows how to shut it down with an alarm after a certain amout of time.
UIf you dont do a request to the session in 60 seconds it shuts down thats why they destroyed it after 60 seconds so you dont get billed for the DO being alive if the session isnt even active anymore
UIf you do a call every 60 seconds their example would live forever
JWhere is it documented that the session shuts down automatically after 60 seconds?
Uhttps://developers.cloudflare.com/browser-rendering/platform/limits/#limits
A browser instance gets killed if it does not get any command for 60 seconds, freeing one instance.
The Workers Browser Rendering API is currently in beta. Add yourself to the waitlist to reserve your spot, and as we expand availability, we will …
J@Unsmart | Tech debt Ah, okay. That's not good at all then.
JSo if you have, in my example a service that gets a request every minute in a half, you'll have to start a new session for each request.
JWhich will end up with a $4k bill each month.
JYou'd want to add a heartbeat request every <60s in order to keep the browser alive to avoid starting new ones. Crazy.
MYep. The problem isn't keeping it alive for a minute but if it's okay to keep it alive for forever. And if so why is that the best practice, seems hacky.
It doesn't make sense that CF needs to charge $0.10 to spin up/spin down a browser rendering instance for 60 seconds but will let you keep it up all month for $0.10 as long as you interact with it once every 60 seconds. The economics just don't make sense.
It doesn't make sense that CF needs to charge $0.10 to spin up/spin down a browser rendering instance for 60 seconds but will let you keep it up all month for $0.10 as long as you interact with it once every 60 seconds. The economics just don't make sense.
JAgreed. Unless it's basically free for them to have it sit idle... but if that's the case, why automatically turn them off after 1 minute?
MYeah or just make the session pricing low enough so that it makes sense to spin up/spin down as needed rather than relying on hacks to avoid a $4k bill.
JDo CF devs monitor this channel? Or is it only community members speculating?
AIt's a very puzzling situation. I'm basically waiting until it's sorted out before I implement anything with the Browser Rendering API
HHey there when using the Browser Rendering API, I ran into a cloudflare challenge on a website I wanted to access lol. Is there anyway to wait/bypass that? Or is this a goner? Thanks!
HHey Dan did stealth work out for you?
DI haven't tried it, I don't think anyone answered my question on pricing so I didn't try it
DI think that's the point, they don't want bots
HThat's a bummer... will probably have to work a way around then. Thanks for answering bro!
VIs this supposed to work on Pages function?
