Releases take a little to go out, hotfix is up - should be fixed in ~30 mins
Releases take a little to go out, hotfix is up - should be fixed in ~30 mins
Ksame
K
JYep same
W^
Wfix should be out in 5-10
WFix is rolled out
WConfirmed settings load again, please reload the dash and try again
JWorking now
WThanks for raising and pinging!
SIs there a way to publish a page to prod from a preview branch?
We sometimes run into the issue that the prod-build was successful but the page has problems anyway and requires a re-build.
Would be really helpful to test a prod build first before releasing it.
We sometimes run into the issue that the prod-build was successful but the page has problems anyway and requires a re-build.
Would be really helpful to test a prod build first before releasing it.
SI created a thread in pages-help, thanks
Are: Images and CF Pages
I'm having a devil of a time trying to understand this. Any thoughts are very much appreciated.
*The situation: *
I currently plan to deploy a three-page site built with Remix to Cloudflare Pages. One of the pages has about 20 images on it. None are fixed size. They scale with the viewport. The layout changes responsively based on media queries (e.g., one column, two column, three column).
The client doesn't have deep pockets, naturally. I plan to handle transcodes and srcset and sizes attributes for these images on my own.
I'm struggling to understand delivery/storage costs, though.
Ideas:
I'm having a devil of a time trying to understand this. Any thoughts are very much appreciated.
*The situation: *
I currently plan to deploy a three-page site built with Remix to Cloudflare Pages. One of the pages has about 20 images on it. None are fixed size. They scale with the viewport. The layout changes responsively based on media queries (e.g., one column, two column, three column).
The client doesn't have deep pockets, naturally. I plan to handle transcodes and srcset and sizes attributes for these images on my own.
I'm struggling to understand delivery/storage costs, though.
Ideas:
- CF Cache. My original plan was to host the images in the project repo. At some point, some docs lead me to believe that, when a request comes in for an image on a CF-managed site, the image would get copied into cache and would then be served from there until expiration (which I'd love to set to a year?).
- This is my favorite option. If true, how do I calculate pricing?
- R2 w/ or w/o Cache. I see an alternate plan might be a R2 bucket with no need for image transformation. What would the costs be in this scenario? I guess: (a) storage and (b) bandwidth for delivery? I'm super-unclear about how to think about (b). I see discussion about no egress fees, but I'm not sure whether egress and image-delivery bandwidth are the same?
- Can this be paired with Cache, too? How is pricing calculated, if so?
- Images. Finally, I've read about Images, but am not clear if I could use this service for, basically, delivery only. In the situation I've described, is there some way to use this service effectively? And how do I assess its cost?
HIf those images are just used on the page, and you aren't using them primarily in other applications, then you should be fine to just host them in the repo directly. Then, even if they aren't served from cache, they will not invoke billable requests
HUnless Remix decides to serve the images for you. You should be able to tell it not to, if it does
AYeah, just on the page. It's a nice site, but just a site. That's kinda amazing. No fuss, no muss.
- Is CF likely to copy them into its cache? Just trying to sort out my mental model for all this.
- And would GH/Cache serving be terribly slow, i.e., is there a downside I'm not considering?
A(And thanks much for the thoughts.)
HSort of? Files on Github/lab are run through the build process, if there is one, then uploaded to KV. Pages will then, when a request is received, pull the assets from one of the KV origins(EU or NA), cache it in the datacenter it was requested from, and then serve it to the user. Whether the request was cached or not, it will be served at no charge to you, unless the route that was invoked also ran a Function(in this case, via Remix).
AAh. OK.
So I guess there's an intermediate layer in CF which is serving as a kind of cache (KV), even if it isn't some official thing designated as "Cache."
And I suppose this must be part of the overall Pages service you're signing up for when you use it to deliver a site...?
This was very helpful. Thanks. I may get a little confused again between now and final deploy, but I feel better about this now — seems like I have a viable plan.
So I guess there's an intermediate layer in CF which is serving as a kind of cache (KV), even if it isn't some official thing designated as "Cache."
And I suppose this must be part of the overall Pages service you're signing up for when you use it to deliver a site...?
This was very helpful. Thanks. I may get a little confused again between now and final deploy, but I feel better about this now — seems like I have a viable plan.
HNot quite? In this case, KV is the backing store and the cache for Pages
HOnce a deployment is created, it is completely independent of Github. In fact, you should be able to delete the repo entirely, and the Pages Project should continue to work uninterrupted
AOK. That's a good place for me to start. I'll probably circle back as I get closer to finishing.
Really appreciate it.
Really appreciate it.
J
J
JFor orgs it should be:
https://github.com/organizations/{ORG_NAME}/settings/installations/{ID}
Nis there a max request duration for a pages function
Nor like a way to see if a request is timing out
Nin the logs
Nwould a 500 code mean its timing out
Nbecause it makes no sense... i have a log, a fetch, and a log
Ni see the first log
Nnot the second
Nand it returns 500
Nnvm im stupid
Ni missed the error log
Nalso noticed that the logging solution we use (axiom) released an update 8 years ago adding 
cache: 'no-cache'
WFree plan: 10ms CPU time
Paid plan: 30 seconds of CPU time
Paid plan: 30 seconds of CPU time
W500 can mean a million things haha, best to look at lots
Nwas wondering about overall time
Nlike could i wait for a db response for 10 minutes lol
WYep
WBut your eyeball may disconnect before then
WCloudflare zones have a general read timeout of 100 seconds
WWhich can be adjusted on ENT
WBut you could stream for hours if you wanted
WYeah we don't have info if its a user or org
WIt's been a bug since the start of pages
Jah ok, could potentially have two links there, one formatted for org and one formatted for the user. That should solve it for now as you should have all that information available to you