Cloudflare

my bet is the custom domain caching is taking over

hmm, I thought I lost some features by doing that

right, won't I lose header transforms and such?

cool yeah I just did a Transform Header rule instead

question, why does the main SPA html page have caching disabled completely?

```
< cache-control: public, max-age=0, must-revalidate
```


https://cdn.discordapp.com/attachments/789155108529111069/1112418239994150942/image.png

im just doing this stuff in my front end code
https://cdn.discordapp.com/attachments/789155108529111069/1112418373301698670/image.png

```
const showPaywall = computed(() => {
  if (typeof window === 'undefined') {
    return
  }
```

looks like you have some build-time static rendering going on, are you using Nuxt?

makes sense, vitepress is running your code to generate HTML at build time

you can use this: https://vitepress.dev/guide/ssr-compat#clientonly

Is it possible/easy to create a staging environment with Pages? E.g. basically `main` except using preview or a different set of Env vars?

It is pretty much baked in, you can set all your variables in the settings of the project for production or any other branch, it will build every commit from every branch of you want it to

Well there are some things that don't lend themselves too well to a specific staging environment IMO

Eg. Environment vars can only be set for production, or preview environments. You can't designate a specific preview as "staging" and have different env vars

Same with compatibility flags and bindings

So you might prefer to create another pages site and use it's production branch as "staging"

Yah, that's the rub. Would like it to be independent of preview and prod, but I might just have it be the same as preview for now

Ohh, 
> process.env. doesn't exist in Cloudflare workers
I just need to use it directly, like `EXAMPLE`. Unconventional

I don’t see how it’d exist directly either

The build environment is a Ubuntu container, that runs Node

Workers and Functions runtime aren’t Node, they’re a custom runtime, and variables are provided to the fetch handler

Oh, I guess it wouldn't work

Im just trying to get my environment variables in the build process, which for me is node 21. So I'm still not sure why `process.env.EXAMPLE` is undefined

Do you see the default ones, like CF_PAGES_BRANCH?

Nope, that's undefined too. I ran:
```
console.error(`process.env.CF_PAGES_BRANCH: ${process.env.CF_PAGES_BRANCH}`);
console.error(`CF_PAGES_BRANCH: ${CF_PAGES_BRANCH}`);
```
Which gives me:
``` process.env.CF_PAGES_BRANCH: undefined
ReferenceError: CF_PAGES_BRANCH is not defined
```

I think i've found a bug in turbo... This doesn't work:
```
"env": ["SENTRY_AUTH_TOKEN"],
```

I need to put it in 
```
"globalEnv": ["SENTRY_AUTH_TOKEN"],
```

Yea, that's fixed it :), thanks for your replies

If so, see how I add env variables here (on hono discord)
https://discord.com/channels/1011308539819597844/1012485912409690122/1181284862683009094

Nope I use tRPC running on Node (not on cloudflare workers). I only have a cloudflare pages frontend running on cloudflare pages

I'll still take a look, thanks for the link

Why would `process.env.EXAMPLE` be

cvf-2x

Oh weird, its happening way more frequently for firefox:

It still happens on chrome, but much less frequently (left chrome, right firefox)

I noticed they were cycling back and forth through a limited selection, it wasn't just random

Weirdly, this has stopped happening now 🤔. No changes from my side

Might be related to deployment, where CF pages doesn't always serve the latest build

image

that's pretty typical cache behavior, if you're hitting cloudflare location which is called a MCP (multi-colo point of presence), its a location/city which is served via mutiple cloudflare datacenters/colos. Each colo has its own cache, and when your connection dies you get routed randomly.
(so one has one version in cache, another has another, etc. Some browsers end the connection when you hard refresh)

You shouldn't have Cache in front of pages at all though (Cloudflare unfortunely has it on for default for specific asset types). Pages has its own internal cache, not reflected by the headers. You can make a simple cache rule to disable it, ex:

You can find cache rules under Cache -> Cache Rules

Oh wow, lovely, thanks. I'll give that a go. Great idea to check the response headers, didn't think of that 🥹

Does anyone know why file hashes are

Sorry for the tag, <@905865275583365161> I see you have been helping many others with the same issue. please have a look at your convenience. 🙏🏻 
- pages URL (working): https://forest-landing.pages.dev
- custom domain (not working): https://forest.chainsafe.io/

ERR_SSL_VERSION_OR_CIPHER_MISMATCH means there's no ssl cert issued

Check over https://developers.cloudflare.com/pages/configuration/debugging-pages/#missing-caa-records
Specifically the CAA part

> ;; ANSWER SECTION:
> chainsafe.io.           300     IN      CAA     0 issue "comodoca.com"
> chainsafe.io.           300     IN      CAA     0 issue "digicert.com"
> chainsafe.io.           300     IN      CAA     0 issue "letsencrypt.org"
> chainsafe.io.           300     IN      CAA     0 issuewild "amazon.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "amazonaws.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "amazontrust.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "awstrust.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "comodoca.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "digicert.com"
> chainsafe.io.           300     IN      CAA     0 issuewild "letsencrypt.org"
You're missing pki.goog

Awesome, thanks for the quick check, I will add the CAA and see.

<@905865275583365161> a quick question please, should I add all 8 CAA records for my subdomain `forest.chainsafe.io`? or just adding the missing `pki.goog` for the root domain `chainsafe.io` is enough?

You've already got all of the other ones as far asI can  see. Pages itself only uses Let's Encrypt or GTS/pki.goog

oh you mean specifically for that subdomain, CAA works recursively, since there's nothing on forest it uses the ones on chainsafe.io

So you can just add pki.goog to root. I would then wait a few mins and readd the custom domain and see

ok got it, I will add below two missing CAAs for root domain, and expect the custom domain `forest.chainsafe.io` gets SSL cert.
1. `CAA chainsafe.io  issue pki.goog` 
2. `CAA chainsafe.io  issuewild pki.goog`

You don't need the issuewild but it won't hurt. I would also remove and readd the pages Custom domain the Pages Custom Domain tab, otherwise, if it's been more then 7 days it's already given up, even if it hasn't been it's still way slower as it exponentially falls off in retry intervals overtime: https://developers.cloudflare.com/cloudflare-for-platforms/cloudflare-for-saas/domain-support/hostname-validation/backoff-schedule/
Removing, readding, would reset that

one last question, how do I define this value `cansignhttpexchanges=yes` as I see in the dig output of CAAs? or it is will be add automatically?

don't need it for  Pages. Some DNS Providers may not support it with their editors

you are awesome 🙏🏻 , the pages custom  domain is up
https://forest.chainsafe.io/

Hello I have an issue setting up

This worked before, first happend on 11:43AM January 9, 2024 IST

Can you share deployment IDs of last working and one having errors?

09-Jan-2024 - Failed -  Deployment ID : `2201faa5-9ac3-4c24-ac03-9e76797e3c2b`
08-Jan-2024 - Success -  Deployment ID : `36a418f1-c924-481a-baff-228c4ab182ff`