Hi there, We have a project based on cloudflare workers & queues. We currently have 1 queue with thi

Hi there,
We have a project based on cloudflare workers & queues. We currently have 1 queue with this config ( batch size:50, wait time: 3 sec, msg retries: 1, max concurrency: 10). The workers send batches of messages to the queue and sometimes it gets overloaded thus we expected auto scaling to work, but we've not yet seen auto scaling(concurrency) working there.
For some reason the maximum reference in the metrics graph is 1:
Thanks!

Iiustin.pericica Hi there, We have a project based on cloudflare workers & queues. We currently h...

Matt Silverlock•10/21/23, 3:17 PM

What is your configuration? Do you have a high rate of errors? Concurrency backs off (per the docs) when errors increase.

Old Man Maker•10/22/23, 7:57 AM

I've got a dead basic consumer and I am getting this message:

Unexpected fields found in queues.consumers[0] field: "services"

Unexpected fields found in queues.consumers[0] field: "services"

I am unsure what this is refering to?

Old Man Maker•10/22/23, 7:58 AM

This is the consumer:

async queue(batch: MessageBatch<TransactionModel>, env: Bindings): Promise<void> {
        
        try {
        
            let messages = JSON.stringify(batch.messages);
            console.log(`consumed from our queue: ${JSON.stringify(messages)}`);


        } catch (error) {
            console.log('Error=>',error);
        }
      },

async queue(batch: MessageBatch<TransactionModel>, env: Bindings): Promise<void> {
        
        try {
        
            let messages = JSON.stringify(batch.messages);
            console.log(`consumed from our queue: ${JSON.stringify(messages)}`);


        } catch (error) {
            console.log('Error=>',error);
        }
      },

kian•10/22/23, 9:19 AM

It's in your Wrangler.toml

Kkian It's in your Wrangler.toml

Old Man Maker•10/22/23, 11:41 AM

Yeah - not sure if it should be classified as a bug though? The definition of a service worker anywhere lower in the toml appears to cause these types of error messages. Because mine was below the queue consumer, it looked like it was an error attached to the queue def

Moving it to the top sorted it

kian•10/22/23, 11:45 AM

It’s not a bug, just how TOML works

kian•10/22/23, 11:45 AM

You can change to using [[services]] with keys below it instead of services = [{…}]

Old Man Maker•10/22/23, 11:48 AM

Thanks folks

Original message was deleted

Old Man Maker•10/22/23, 11:51 AM

Correct - it was a perfect storm of mess. Queue defined with correct fields and then a heap of other stuff under it commented out, and then the services definition, - so looking at the toml at the point of the queue looked like it was smoking plastic... scolling down, and it turns out I was smoking plastic

Original message was deleted

Old Man Maker•10/22/23, 11:53 AM

I have a love hate relationship with it at this point.

I kinda wish that every single entry was standardized, but in this case, the flexibility of it, is also the killer. That and the fact I was rushing and not cleaning up my rubbish - self inflicted pain here for sure

Old Man Maker•10/23/23, 10:47 AM

hi folks - I got a worker with a queue consumer, and a binding to a KVNamespace.

I also have this as a HONO app - and I'm not sure if my bindings are actually working at all (is there an easy way to test? they always appear as {} in console.log)

type Bindings = {
    AIS_ALL_TRANSACTIONS:KVNamespace
}


export default {
    fetch: app.fetch,
    async queue(batch: MessageBatch<TransactionModel>, env: Bindings): Promise<void> {

}}

type Bindings = {
    AIS_ALL_TRANSACTIONS:KVNamespace
}


export default {
    fetch: app.fetch,
    async queue(batch: MessageBatch<TransactionModel>, env: Bindings): Promise<void> {

}}

In this case, will the env parameter correctly get the bindings? It does not appear to work for me. I've looked at a few options, but no clearer on a solution.

kian•10/23/23, 10:58 AM

Seems fine to me

Kkian Seems fine to me

Old Man Maker•10/23/23, 11:27 AM

Yeah, I'm a little stumped. Definitely can't read or write to the kv and the dashboard logging is show me all the console logs up to the try/catch error and then no information from the error, so I'm kinda blind here

kian•10/23/23, 11:29 AM

What does your Wrangler.toml look like?

Old Man Maker•10/23/23, 11:31 AM

Normal:

[[kv_namespaces]]
binding = "AIS_ALL_TRANSACTIONS"
id = "xxxxx" <=I have the correct Id in here

[[queues.consumers]]
queue = "my-q" <=Correct queue here as well



# Optional: Configure batching and retries: https://developers.cloudflare.com/queues/learning/batching-retries/
# max_batch_size = 10 //The maximum number of messages to deliver before the timeout
# max_batch_timeout = 30 //default is 5 seconds
# max_retries = 10 //default to 3

# all dead transactions go to the following dead letter queue
dead_letter_queue = "transaction-dead-q"

[[kv_namespaces]]
binding = "AIS_ALL_TRANSACTIONS"
id = "xxxxx" <=I have the correct Id in here

[[queues.consumers]]
queue = "my-q" <=Correct queue here as well



# Optional: Configure batching and retries: https://developers.cloudflare.com/queues/learning/batching-retries/
# max_batch_size = 10 //The maximum number of messages to deliver before the timeout
# max_batch_timeout = 30 //default is 5 seconds
# max_retries = 10 //default to 3

# all dead transactions go to the following dead letter queue
dead_letter_queue = "transaction-dead-q"

Deploy shows its correctly attached

Old Man Maker•10/23/23, 11:34 AM

Is there anyway to get a catch error to show in the dashboard logs?

Old Man Maker•10/23/23, 2:50 PM

Moving this to KV channel as it is not queue specific

Old Man Maker•10/23/23, 3:31 PM

ok - I now understand what is going wrong.
In example code for queues, we often use a foreach loop to work through the batch. This has a nasty bug (the JS side of it) - I remembered coming across a similar issue a while back with async foreach loops in Node.
So, any async work needing to be done within the scope of the queue, you need to use a plain old for loop. This was hours and hours of debugging. I hope no-one else hits this, or at least people will know.

OOld Man Maker ok - I now understand what is going wrong. In example code for queues, we often ...

Unsmart•10/23/23, 3:41 PM

await does nothing on a foreach so this isnt really a bug its expected even if it seems weird: https://stackoverflow.com/questions/37576685/using-async-await-with-a-foreach-loop

Unsmart•10/23/23, 3:44 PM

I believe

Error Lens

Error Lens

is the extension I have on vs code that makes it a bit easier to see these errors for me at least

OOld Man Maker ok - I now understand what is going wrong. In example code for queues, we often ...

John Spurlock•10/23/23, 3:45 PM

you are not the first to encounter this - maybe the cloudflare queue examples should explicitly mention avoiding

.forEach

.forEach

Debugging fetch in a consumer

UUnsmart I believe `Error Lens` is the extension I have on vs code that makes it a bit ea...

Old Man Maker•10/23/23, 4:31 PM

I'll take a look at that extension for sure. Thanks so much

Stefan•10/23/23, 10:30 PM

Hi, i'm currently trying to set up a queue between two different workers, one hosts the main application (the producer), and the other is responsible for firing off verification emails (the consumer), but what i'm not clear about and can't find documentation on, is that, assuming I have to have two wrangler.toml files (which my understanding is that i do), then how do i configure the queue to be exposed to both the consumer and the producer if they are in different files?

SStefan Hi, i'm currently trying to set up a queue between two different workers, one ho...

Old Man Maker•10/24/23, 7:13 AM

You add the producer details to the toml file of the producer workers (you can have multiple) - and you add the consumer details to the toml of the worker that is going to manage the queue.

Stefan•10/24/23, 8:47 AM

Okay so always pair the queue with the consumer, thank you!

�

😈 Donkey 💫•10/25/23, 5:29 AM

Hi everyone,
I have a question:

If:
Client http -> Producer worker -> queue -> consumer worker -> external http service api

How to make producer awaiting, then return response from "external service api" back to client

�

😈 Donkey 💫•10/25/23, 5:29 AM

is that possible?

kian•10/25/23, 5:43 AM

You can’t, no

kian•10/25/23, 5:43 AM

That wouldn’t be a use-case for Queues but rather Service Bindings

CCharlie What kind of message throughput are you atttempting to send here? What's the err...

arbfay•10/25/23, 12:17 PM

I found the problem. The issue was within a DO - I added a

this.env = env

this.env = env

in the constructor to be able to call bindings from methods, then tried to send a last message to a queue right after calling

this.state.storage.deleteAll()

this.state.storage.deleteAll()

, but it turns out calling

deleteAll()

deleteAll()

also deletes the current memory and references of the object (not just storage), so the binding was lost. The error message is a bit confusing though - it made me believe the error was on CF's side (checked status page as well).

bret_pat•10/26/23, 10:22 AM

Are queue consumers going to be using the new "standard" worker pricing model in the future? Or at least, have the option to not be unbound?

Bbret_pat Are queue consumers going to be using the new "standard" worker pricing model in...

Yongyi•10/26/23, 11:40 AM

i am also wondering the same thing

CCharlie @bret_pat @HeartYi consumers for Queues follow the same billing as any other wo...

Yongyi•10/26/23, 4:20 PM

So they are billed based on cpu millisecond time not the wall time?

CCharlie https://blog.cloudflare.com/workers-pricing-scale-to-zero/

omar4289•10/26/23, 10:10 PM

Personally, I just think it got expensive for the unbundled mode (now 20$/1m seconds instead of 12.5$ and 0.3$ instead of 0.15$ per 1m requests). I know in the document we talk about the cpu time, instead of clock time, but to be honest my CPU time is usually very low, and thus the only impacting factor for me would be the number of requests, which unfortunately just doubled in price.

That would not change much for me, because Cloudflare is still a very generous platform and is pretty easy to use, I just don't like how this is promoted as only beneficial

Oomar4289 Personally, I just think it got expensive for the unbundled mode (now 20$/1m sec...

Unsmart•10/26/23, 10:17 PM

~100ms of wall time = + $0.15/million request

Also one unknown part of the unbound billing model is if you used 1 second of wall time and 1 second of CPU time you actually get billed 8 seconds wall time instead. (8x multiplier on CPU time)

So unless you had extremely fast requests it will probably be cheaper or around the same price.

UUnsmart ~100ms of wall time = + $0.15/million request Also one unknown part of the unbo...

omar4289•10/26/23, 10:21 PM

Lets see, but anyway, I still like the products Cloudflare is working ln. Really good job and I am open to pay any extra amount (if any) for the great services you provide at a very reasonable price

omar4289•10/26/23, 10:23 PM

Just one more question, since workers run on a 128MB machines, does it mean that the number of ms that we have is per 1GB of RAM, or is it per worker (128MB)?

Unsmart•10/26/23, 10:25 PM

The unbound billing model is based on 128MB (or 1/8th a GB) so 8 seconds of true wall time = 1 GB-sec for billing

omar4289•10/26/23, 10:30 PM

Ok, but right now, assuming I only run a loop in my program without I/O (very unlikely) and that takes exactly 1second to complete.

In the old model, I would pay 12.5$ for 8 million executions
In the new model, I would pay 20$ for 1 million executions.

I know the I/O work usually takes time, but it all depends on what you do, sometimes it will be cheaper, sometimes more expensive. Lets see how it goes

Unsmart•10/26/23, 10:37 PM

You can use this calculator and input your metrics to see the cost difference: https://calculator.imcf.me/

Unsmart•10/26/23, 10:38 PM

(The calculator doesnt take into account the CPU multiplier so you would need to make sure the wall time ms is at least 8x the cpu time ms)

Mad•10/28/23, 11:58 PM

I have a single queue marked both as a producer and a consumer queue, attached to the same worker. The worker takes input from the queue and produces output to the same queue until the task is done. What I observe is, the task runs for couple of iterations and stops running.

Mad•10/28/23, 11:58 PM

works fine when I run this locally

Mad•10/28/23, 11:59 PM

@Charlie B | KV, Queues, PubSub any idea what must be going wrong. Can I not use the same queue as a producer as well as consumer?

async queue(batch: MessageBatch<TransactionModel>, env: Bindings): Promise<void> { try { let messages = JSON.stringify(batch.messages); console.log(`consumed from our queue: ${JSON.stringify(messages)}`); } catch (error) { console.log('Error=>',error); } },

[[kv_namespaces]] binding = "AIS_ALL_TRANSACTIONS" id = "xxxxx" <=I have the correct Id in here [[queues.consumers]] queue = "my-q" <=Correct queue here as well # Optional: Configure batching and retries: https://developers.cloudflare.com/queues/learning/batching-retries/ # max_batch_size = 10 //The maximum number of messages to deliver before the timeout # max_batch_timeout = 30 //default is 5 seconds # max_retries = 10 //default to 3 # all dead transactions go to the following dead letter queue dead_letter_queue = "transaction-dead-q"

Hi there, We have a project based on cloudflare workers & queues. We currently have 1 queue with thi

Similar Threads

Similar Threads

Similar Threads