Path of least resistance might be to adapt an existing SMTP library to the Worker's TCP API

Xxeon06 Path of least resistance might be to adapt an existing SMTP library to the Worke...

A

ajgeiss0702•10/14/24, 11:55 PM

yeah thats my current plan. Doing some more research to be sure theres not an already existing one for workers

X

xeon06OP•10/14/24, 11:55 PM

That is, overall however, not the easiest solution. That is just switching to any other provider, most of whom have a REST API

X

xeon06OP•10/14/24, 11:56 PM

If you're that set on using your specific SMTP-only provider, another option is hosting an SMTP proxy server on some $5 VPS and then calling that via HTTP from your worker

Xxeon06 If you're that set on using your specific SMTP-only provider, another option is ...

A

ajgeiss0702•10/14/24, 11:58 PM

that also crossed my mine, but im not sure i want to maintain another microservice outside of cloudflare just for proxying stuff from cloudflare. Although I also probably dont want to maintain a smtp library. I'll look at how complicated smtp libraries are and go from there

X

xeon06OP•10/14/24, 11:58 PM

I'd agree with you there. I'm not sure if you have a specific reason for wanting to keep with your current providers, but the options are probably endless, 99% with HTTP APIs

Aajgeiss0702 I have done this. When getting a token, I store it in [KV](https://developers.cl...

H

HS•10/15/24, 12:27 AM

Thanks @ajgeiss0702. Just to confirm, you implemented the cache API on the request being sent to KV, or were you referring to caching the external API request? Any chance you have a code sample that you can share?

A

ajgeiss0702•10/15/24, 12:28 AM

I am putting the token into the cache api, then checking that before checking for the token in KV

A

ajgeiss0702•10/15/24, 12:28 AM

give me a minute to find that code

A

ajgeiss0702•10/15/24, 12:33 AM

ah unfortunately that specific implementation is not open sourced currently, but its basically this for checking using the cache api:

  const cfCache = await platform.caches.open("(cache name here)");

  const cacheMatch = await cfCache.match(cacheURL) as Response | undefined;
  if(cacheMatch) {

    const responseGeneratedRaw = cacheMatch.headers.get("x-response-generated");

    if(responseGeneratedRaw) {
      const responseGenerated = new Date(responseGeneratedRaw);

      if(Date.now() - responseGenerated.getTime() < cache_time) {
        // return data from cache match
      }
    }
  }

  const cfCache = await platform.caches.open("(cache name here)");

  const cacheMatch = await cfCache.match(cacheURL) as Response | undefined;
  if(cacheMatch) {

    const responseGeneratedRaw = cacheMatch.headers.get("x-response-generated");

    if(responseGeneratedRaw) {
      const responseGenerated = new Date(responseGeneratedRaw);

      if(Date.now() - responseGenerated.getTime() < cache_time) {
        // return data from cache match
      }
    }
  }

  const cfCache = await platform.caches.open("(cache name here)");

  const cacheMatch = await cfCache.match(cacheURL) as Response | undefined;
  if(cacheMatch) {

    const responseGeneratedRaw = cacheMatch.headers.get("x-response-generated");

    if(responseGeneratedRaw) {
      const responseGenerated = new Date(responseGeneratedRaw);

      if(Date.now() - responseGenerated.getTime() < cache_time) {
        // return data from cache match
      }
    }
  }

  const cfCache = await platform.caches.open("(cache name here)");

  const cacheMatch = await cfCache.match(cacheURL) as Response | undefined;
  if(cacheMatch) {

    const responseGeneratedRaw = cacheMatch.headers.get("x-response-generated");

    if(responseGeneratedRaw) {
      const responseGenerated = new Date(responseGeneratedRaw);

      if(Date.now() - responseGenerated.getTime() < cache_time) {
        // return data from cache match
      }
    }
  }

and for storing the response in the cache api:

platform?.context?.waitUntil(
  cfCache.put(cacheURL, await createMFResponse(jsonResponse.clone()) as Response)
)

platform?.context?.waitUntil(
  cfCache.put(cacheURL, await createMFResponse(jsonResponse.clone()) as Response)
)

platform?.context?.waitUntil(
  cfCache.put(cacheURL, await createMFResponse(jsonResponse.clone()) as Response)
)

platform?.context?.waitUntil(
  cfCache.put(cacheURL, await createMFResponse(jsonResponse.clone()) as Response)
)

theres some conversion stuff there because miniflare's response object doesnt match the web api, but you might not have to deal with that depending on what youre doing

Aajgeiss0702 ah unfortunately that specific implementation is not open sourced currently, but...

H

HS•10/15/24, 1:04 AM

@ajgeiss0702 Thank you so much. I'll try to follow this approach for my use-case. Really appreciate the help!

S

shun•10/15/24, 1:18 AM

Rebuild feature is very helpfull

P

Pato•10/15/24, 1:37 AM

hey! does anyone knows why when testing a service worker in localhost gives me this error:

TypeError: Cannot read properties of undefined (reading 'fetch')

TypeError: Cannot read properties of undefined (reading 'fetch')

TypeError: Cannot read properties of undefined (reading 'fetch')

TypeError: Cannot read properties of undefined (reading 'fetch')

but when deploying the exact same code to a worker and testing it, works fine? I

ve been having this issue for a while now. 
My wrangler version:

ve been having this issue for a while now. 
My wrangler version:

ve been having this issue for a while now. 
My wrangler version:

ve been having this issue for a while now. 
My wrangler version:

wrangler 3.4.0`

S

shun•10/15/24, 2:36 AM

I don't want to have sensitive information (api key, etc.) in wrangler.toml, so I am running a shell in the build command that dynamically creates wrangler.toml from the build variables and runs wrangler deploy using that wrangler.toml.

How does everyone protect their sensitive information?

Sshun I don't want to have sensitive information (api key, etc.) in wrangler.toml, so ...

H

Hello, I’m Allie!•10/15/24, 2:50 AM

You shouldn’t be putting sensitive info in your wrangler.toml. Secrets can be deployed separately

HHello, I’m Allie!You shouldn’t be putting sensitive info in your wrangler.toml. Secrets can be de...

S

shun•10/15/24, 2:54 AM

Thank you very much! Yes, I know. So I have created wrangler.toml.template and run the build command to generate the wrangler.toml file shell with the build variables.
You can also refer to the values registered in github actions secrets in the secrets property of github actions, but I think it will be difficult to manage if you register PRD_API_KEY, STG_API_KEY, etc. for each environment. So I have Cloudflare Workers (PRD, STG, E2E, Test, etc...) created for each environment. I register environment variables in the build variables of What do you think about this solution?

Sshun Thank you very much! Yes, I know. So I have created wrangler.toml.template and r...

H

Hello, I’m Allie!•10/15/24, 3:31 AM

Did you do a

wrangler secret put

wrangler secret put

wrangler secret put

wrangler secret put for the deploy?

HHello, I’m Allie!Did you do a `wrangler secret put` for the deploy?

S

shun•10/15/24, 4:38 AM

OH, I didn't know there was such a practice! I will try it!

B

B1397KB•10/15/24, 6:23 AM

Hello everyone, how do you handle the long TCP SSL handshake when the worker is performing a fetch operation?

B

B1397KB•10/15/24, 6:24 AM

Is there any way to reduce the handshake time? In the worker's fetch

B

B1397KB•10/15/24, 6:25 AM

B

BASIS-64 Computer•10/15/24, 6:59 AM

Hi

B

BASIS-64 Computer•10/15/24, 6:59 AM

will i get charged if cloudflare workers free plans reached the limit (100000 requests/day) ?

B

Brett Willis•10/15/24, 7:37 AM

We have an API which ingests images from the user and stores them in an R2 bucket. One part of that API which is hard to implement on the Workers platform is image transformation and validation. Use-cases such as

transforming the image on ingest to compress and shrink to a certain max resolution
validating image dimensions (must be 128x128px for example)
Does anyone have any experience or recommendations with these use-cases?
We used
```
sharp
```
sharp
```
sharp
```
sharp
which cannot run on Workers. Perhaps could use
```
jimp
```
jimp
```
jimp
```
jimp
but it's probably costly on CPU and Workers have limited memory. Cloudflare Images seems to be more about serving images with transformations, rather than transforming/validating on ingest via a worker (would prefer to store a single minified image in R2)?

0

0x146231489231923•10/15/24, 7:44 AM

how do yall do logging?

0

0x146231489231923•10/15/24, 7:44 AM

in cloudflare worker

BBASIS-64 Computer will i get charged if cloudflare workers free plans reached the limit (100000 re...

D

D Trombett•10/15/24, 7:51 AM

If you didn't upgrade to a paid plan no, further requests will simply fail

M

masd4•10/15/24, 9:43 AM

Hi, the feature matrix for "workers vs. pages" says that middleware isn't supported and to use service bindings. But how does that help to secure a *.workers.dev Url? That's an url I don't control, but I still want basic auth for those...

W

Walshy•10/15/24, 9:45 AM

We recommend disabling workers.dev or putting Access on it

M

masd4•10/15/24, 9:46 AM

I'm trying to achieve that somehow but the options seem very limited because "Workers which have static assets cannot be routed on a URL which has a path component".. so essentially I would need an entire custom domain just to have my worker for assets available anywhere. Half the problem might be that Cloudflare Access is probably overkill even if I do get access for that on the company account

PPato hey! does anyone knows why when testing a service worker in localhost gives me t...

A

Ashley•10/15/24, 10:55 AM

Your Wrangler version is very out of date - I had a similar problem, I think even now local service bindings are experimental - but I’d try updating to the latest version of Wrangler as a first step

Original message was deleted

H

Hello, I’m Allie!•10/15/24, 1:36 PM

Using a Router?

Original message was deleted

H

Hello, I’m Allie!•10/15/24, 1:37 PM

Any router

H

Hello, I’m Allie!•10/15/24, 1:37 PM

At least I know Hono and Itty could do that

A

Aster•10/15/24, 2:05 PM

Hey!

I am looking to expose prometheus metrics out of my workers.

What could be the best way for this?

have DO that stores the metrics. And expose the metrics on another /metrics worker
use analytics engine to store the metrics, and expose them on another /metrics worker
build recorded queries on top of logs with something like Loki?

What would you suggest

?

A

Aster•10/15/24, 2:08 PM

Basically I want to end up with otel metrics like:

http_server_duration_count{worker="my-worker", http_status_code=~"2.."}

http_server_duration_count{worker="my-worker", http_status_code=~"2.."}

http_server_duration_count{worker="my-worker", http_status_code=~"2.."}

http_server_duration_count{worker="my-worker", http_status_code=~"2.."}

W

Walshy•10/15/24, 2:13 PM

i love Analytics Engine, prefer that over doing Prometheus in a DO

W

Walshy•10/15/24, 2:13 PM

but you can hack promjs to run in a DO and serve - others have done this before

A

Aster•10/15/24, 2:14 PM

Why would you prefer AE over Prom in a DO?

A

Aster•10/15/24, 2:15 PM

and I guess Prom in a DO means saving the state of a counter in the DO

0

0x146231489231923•10/15/24, 2:22 PM

how you all do logging for production debugging in worker?

0

0x146231489231923•10/15/24, 2:22 PM

the free plan doesnt do much

A

ac•10/15/24, 2:31 PM

Is there an issue with worker CI/CD builds? This happened yesterday and it's happening to me again, but all of my workers build statuses are just "Queued" even though there are no builds running.

A

ac•10/15/24, 2:43 PM

^ I believe I've found a bug with Workers CI/CD, what's the best place to report that? (I was able to reproduce the issue, and it seems related to my setup and how Workers CI is triggered)

C

Constantin•10/15/24, 2:56 PM

can anyone please advice related to custom domains. I'm trying to create a sass king of app.

I managed to create tenant wildcard subdomains by using pages and a custom proxy worker.

eg. https://ionica.app.greatoutdoors.live/

I added the custom domain on my side
tenant added CNAME servicii.cmnstmntm.com -> ionica.app.greatoutdoors.live

however, the request to https://servicii.cmnstmntmn.com/ is timing out. this is the traceroute

C

Constantin•10/15/24, 2:57 PM

is this even possible with my plan? i assumed it is since i could add the custom domain

CConstantin can anyone please advice related to custom domains. I'm trying to create a sass ...

C

Chaika•10/15/24, 3:02 PM

Do you have a wildcard catch-all route? https://developers.cloudflare.com/cloudflare-for-platforms/cloudflare-for-saas/start/advanced-settings/worker-as-origin/.
Unproxied CF For SaaS hostnames come through as the actual hostname, ex it's looking for a worker route matching

servicii.cmnstmntmn.com

servicii.cmnstmntmn.com

servicii.cmnstmntmn.com

servicii.cmnstmntmn.com and Worker Custom Domains just won't work
Be careful not to break your existing subdomains/domain with that. I usually recommend people get a separate domain just for customer custom domains to avoid issues with that and the fact that all custom hostnames inherit your zone's configuration (waf rules, config rules, network settings, etc), so better to eliminate the possibility of non-intentionally changing settings for them/more flexibility with your custom domains

Cloudflare Docs

Workers as your fallback origin | Cloudflare for Platforms docs

Learn how to use a Worker as the fallback origin for your SaaS zone.

C

Constantin•10/15/24, 3:08 PM

yes, the worker has a pattern like this

C

Constantin•10/15/24, 3:14 PM

so i need a worker that handles the origin specifically

C

Constantin•10/15/24, 3:15 PM

thanks @Chaika i'll give it a try

CConstantin yes, the worker has a pattern like this

C

Chaika•10/15/24, 3:24 PM

no that's not a wildcard catch-all (well, not as wide as CF For SaaS needs)