I’m using D1, so not sure I can use alarms. Until unless I implement some code which stores in durab

I’m using D1, so not sure I can use alarms. Until unless I implement some code which stores in durable objects and wake up at certain time and then pushes the data to d1.

Mmau I’m using D1, so not sure I can use alarms. Until unless I implement some code w...

Hard@Work•8/6/24, 10:53 AM

Create a DO for each notification the user sets, then set an alarm

mauOP•8/6/24, 10:58 AM

Just to clarify… you mean one do object per notification and then set the alarm?

mauOP•8/6/24, 10:58 AM

Is there is any limit on alarm time limits?

Mmau Just to clarify… you mean one do object per notification and then set the alarm?

Hard@Work•8/6/24, 11:19 AM

That is correct. As for limits for the time, as long as it is representable, it should be able to fire

mauOP•8/6/24, 11:20 AM

I will give it a try. Thanks for the solution

Dlanda Griffin•8/8/24, 4:17 PM

dsds10•8/8/24, 6:11 PM

Can somebody explain to me why there is a generous limit of 10k queues per account? Maybe I‘m missing something obvious, but only pull is supported over HTTP, push would still need a worker binding (there is a max of 500 workers per account, although you can bind multiple queues to a single worker iirc). The only explanation would be workers for platforms, but then kv namespace and r2 bucket limits won‘t match that idea

Ddsds10 Can somebody explain to me why there is a generous limit of 10k queues per accou...

Matt Silverlock•8/8/24, 8:52 PM

Many folks have more than one script and Workers for Platforms use-cases. One Worker script can effectively bind ~5000 or so resources. Other products have different metadata constraints.

MMatt Silverlock Many folks have more than one script _and_ Workers for Platforms use-cases. One ...

dsds10•8/9/24, 12:07 PM

got it thx

flyingshoes•8/12/24, 8:42 PM

Is there a limit to how many tasks I can push into a queue? I'm thinking of having a cron job push tasks into a queue, and the limit could be pretty high (potentially millions). would the limit be the 1200 requests/ 5 minutes?

Fflyingshoes Is there a limit to how many tasks I can push into a queue? I'm thinking of havi...

John Spurlock•8/12/24, 8:42 PM

the limits are documented here: https://developers.cloudflare.com/queues/platform/limits/

Cloudflare Docs

Limits · Cloudflare Queues

Many of these limits will increase during Queues’ public beta period. Follow our changelog or join the #queues-beta channel in our Developer Discord …

KKam I'm seeing queue `send` and `sendBatch` calls taking anywhere from ~250ms to ove...

Brett Willis•8/12/24, 8:47 PM

Can't comment on those latencies myself, but you could consider using

ctx.waitUntil()

ctx.waitUntil()

to more cheaply and easily defer those operations asynchronously without holding back the response.

JJohn Spurlock the limits are documented here: https://developers.cloudflare.com/queues/platfor...

flyingshoes•8/12/24, 8:58 PM

got it -- so 400 messages per second. thank you.

KKam I'm seeing queue `send` and `sendBatch` calls taking anywhere from ~250ms to ove...

Pranshu Maheshwari•8/12/24, 10:13 PM

it's a bit higher than expected, but not out of the ordinary. We're working on improvements that should improve latency for sending messages to a queue. Those improvements should ship by Oct.

KKam Thanks for the info Pranshu! Any idea what latency you are targeting with the im...

Pranshu Maheshwari•8/13/24, 3:11 PM

It's too soon to share specific targets, but stay tuned!

NoGameNoFun•8/14/24, 3:26 AM

Hi everyone, I like the worker and queue features in cloudflare, and been trying out about few weeks already.

I got a question, I noticed the consumer for the queue, after some time (like 2 days), it will stop processing the queue message. (the backlog size increase and doesnt get consume)

I attached the screenshot on the consumer setting, is this a bug or is just because of my consumer setting?

jordanfinners•8/14/24, 9:13 PM

Please can I provide some feedback/feature requests for cloudflare queues

Firstly the dev experience inside workers is pretty great! It would be awesome if it could work with Pages functions so I don't have to deploy separate consumers but thats a nit really.

The metrics in the dashboard are useful but its missing the main one for me, how many messages are in the queue at that point in time, especially important for deadletters
Redriving of deadletter messages, either a button in the UI or a way of doing it via HTTP or JS.
Currently you can pull the messages from the queue via HTTP but can't acknowledge them or push back into the queue locally, meaning I would need to deploy a worker just to redrive each queue which seems excessive and I need to control what I'm redriving or not.
I had to do 200 messages manually this morning
Alerts on the queue length, ie. If messages drop into a deadletter
Would be helpful if the error message was attached to the queue message in some way like SQS does
Being able to send multiple messages via the UI either file upload or pasting in
Documentation is missing for the Preview endpoint in the Queue REST API docs
Its only happened once but I've had one message not get sent, even though it said it has, which worries me slightly.
If possible increase the batch size, up to 200 would be helpful
Alongside this if possible to increase the timeout period as well, I'm running async reports and these can take several minutes to process and so they are timing out/hitting worker limits.

Happy to talk this through or provide more info if

Wwindygod9986 Does anyone know if cloudflare has any plans to support queue consumers on page...

Matt Silverlock•8/17/24, 12:38 AM

Not in the near term. Pages doesn’t have the concept of any non-request handler (cron, queues, tail)

lulupineapple•8/18/24, 2:06 PM

Hi, how do I deal with queue consumer timeout issue? i.e. make sure queue consumer completed the task with payload? Is there anyway to have queue retry calling consumer unless consumer explicitly acknowledge the payload was processed susessfully?

oof2win2•8/21/24, 12:21 PM

how can i set proper types for a queue?

oof2win2•8/21/24, 12:23 PM

like it seems that i need to change it on the generated env types but then i can't have them be generated

oof2win2•8/21/24, 12:45 PM

more importantly it doesn't work for receiving the messages via the same queue

Ooof2win2 more importantly it doesn't work for receiving the messages via the same queue

Matt Silverlock•8/21/24, 10:41 PM

The TypeScript types for the

queue

queue

handler (your consumer) accept a type parameter - so

async queue<T>(batch: MessageBatch, env: Env)

async queue<T>(batch: MessageBatch, env: Env)

- which will mean that each message in

MessageBatch

MessageBatch

has a typed payload based on your type parameter

oof2win2•8/22/24, 5:55 AM

thanks!

snapo•8/22/24, 7:30 AM

Hey people!

When developing locally, I have configured my queue with following retry timeouts and counts:

max_retries = 10,  
retry_delay = 30

max_retries = 10,  
retry_delay = 30

I noticed that for some unknown reason, the retry logic only attempts up to 3 times. If I lower the retry_delay to 10 seconds, I can reach 6th attempt. If I finally lower the delay to 2 seconds, I can reach 10 retry attempts which is expected.

Any ideas, why the 30s delay is not working and 2 seconds is?

Kaique Nunes•8/22/24, 10:31 AM

it's possible to list messages inside dead queue?

KKaique Nunes it's possible to list messages inside dead queue?

snapo•8/22/24, 11:34 AM

You can do it via the dashboard. If you want to use http then you could configure a pull consumer for the dlq

Ssnapo You can do it via the dashboard. If you want to use http then you could configur...

Kaique Nunes•8/22/24, 11:59 AM

So I would need to implement my own api to retrieve the dlq messages?

snapo•8/22/24, 12:00 PM

No no. See here: https://developers.cloudflare.com/queues/configuration/pull-consumers/

Cloudflare Docs

Pull consumers | Cloudflare Queues

A pull-based consumer allows you to pull from a queue over HTTP from any environment and/or programming language outside of Cloudflare Workers. A pull-based consumer can be useful when your message consumption rate is limited by upstream infrastructure or long-running tasks.

snapo•8/22/24, 12:01 PM

https://developers.cloudflare.com/api/operations/queue-v2-messages-pull

Cloudflare API Documentation

Interact with Cloudflare's products and services via the Cloudflare API

Kaique Nunes•8/22/24, 12:08 PM

Thanks

KKaique Nunes it's possible to list messages inside dead queue?

Pranshu Maheshwari•8/23/24, 2:35 PM

Dead letter queues behave just like normal queues. You could setup a consumer, just like you would with a normal queue. It doesn't necessarily need to be a pull consumer; you could use a worker too.

PPranshu Maheshwari Dead letter queues behave just like normal queues. You could setup a consumer, j...

Kaique Nunes•8/23/24, 2:46 PM

But I would need to store the messages somewhere

Kaique Nunes•8/23/24, 2:47 PM

I don’t want to need to add an extra db only for this

Kaique Nunes•8/23/24, 2:47 PM

Ideally queues could allows us to list pending messages

piffie•8/25/24, 9:56 PM

is there a way of getting the name of the queue, when you have the queue object?

clearwater4227•8/25/24, 9:57 PM

Hi everyone! I am starting looking into using queues but have one question...can I configure a worker script as a queue consumer through terraform? I havent found the setup yet..still looking but if anyone here knows or can point me to the info...

clearwater4227•8/25/24, 9:58 PM

it looks like I can configure the queue reference on a worker to send a message but I need to configure the consumer through terraform also.

clearwater4227•8/25/24, 10:04 PM

or is it just a worker with the queue() function? hmm...Ill try it

clearwater4227•8/25/24, 10:21 PM

hmmm...its not picking up my worker as a consumer... is there a way to configure a queue consumer through terraform?

piffie•8/25/24, 10:51 PM

never used terraform for it - sorry

berkinovish•8/26/24, 8:42 AM

Is there a way to force a queue to increase concurrency, or to set a minimum concurrency?

I'm seeing the following behaviour now:

I have a queue where only once every 1-5 minutes, a message is queued. A message takes about 30 seconds wall time to be processed. Maximum consumer concurrency is on default.

But, when a message comes in that all of a sudden takes 30 minutes to process, the whole queue will be stuck until that one message is processed (that's the behaviour I'm observing at least). This behaviour is in line with the docs, since Queues will only check after processing a batch of messages if the concurrent consumers should be adjusted.

Very annoying behaviour of course - because of a slow message, I suddenly have messages waiting for 30 minutes in the queue. Only after the message is processed, the concurrency will be updated and all waiting messages will be handled. Being able to set a minimum concurrency would be a big help here

John Spurlock•8/26/24, 4:17 PM

anyone else see queue send failures across of dozens of colos at around 2024-08-26 06:21:46 UTC?

Error: Queue send failed: Internal Server Error

Error: Queue send failed: Internal Server Error

Haven't seen anything across so many colos at once like this before

JJohn Spurlock anyone else see queue send failures across of dozens of colos at around 2024-08-...

Matt Silverlock•8/26/24, 4:42 PM

@Pranshu Maheshwari

JJohn Spurlock anyone else see queue send failures across of dozens of colos at around 2024-08-...

Matt Silverlock•8/26/24, 4:45 PM

FWIW I don't see anything in top-level metrics for Queues — no meaningful dip in write or read throughput around 6:21 UTC or increase in 5xx error ratio. Pranshu can dig in with the team.

I’m using D1, so not sure I can use alarms. Until unless I implement some code which stores in durab

Similar Threads