Is there any news on having singleton workflows / limiting a specifically workflow to concurrency of

Is there any news on having singleton workflows / limiting a specifically workflow to concurrency of 1?

Hey folks, I’m just getting started with Workflows and ran into something while testing locally. I noticed that

instance.terminate()

instance.terminate()

isn’t implemented in local dev. I figured restarting the server would clear everything, but even after a restart the workflow instance still seems to persist.

Does anyone know where the instance state is actually stored in local mode? And what’s the recommended way to create a new instance with the same ID if the previous one errored out and I want to retry with different input (since I can’t cancel or terminate locally)?

d3v1an7•9/15/25, 6:49 AM

Hey there, I have a quick question about Workflow limits

From limits docs: https://developers.cloudflare.com/workflows/reference/limits/

Concurrent Workflow instances (executions) per account: 4500
Maximum number of queued instances: 100,000

For queued instances, the description is: "the Workflow is queued, but not currently running. This can happen when you are at the concurrency limit and new instances are waiting for currently running instances to complete."

Question: Does a workflow that is currently in a "Sleep" step count as being active/concurrent? Queued? Something else?

Usage: I'd like to make a workflow that creates a user, then waits for a few days to check if the user has signed in.
This will lead to many "active" workflows that are mostly on a "sleep" step, and I can't work out from the docs if I'm going to run into limit issues or not.

Ty!

Vvijayaraagavan Hey folks, I’m just getting started with Workflows and ran into something while ...

d3v1an7•9/15/25, 6:54 AM

By default, running

wrangler dev

wrangler dev

will create a

.wrangler/state

.wrangler/state

directory where the command was run. You can also change this if you'd like!

Dd3v1an7 Hey there, I have a quick question about Workflow limits ✨ From limits docs: h...

avenceslau•9/15/25, 8:22 AM

When a workflow goes to sleep they stop counting towards concurrency limits and are counted as queued again

Vvijayaraagavan Hey folks, I’m just getting started with Workflows and ran into something while ...

Olga Silva•9/15/25, 10:23 AM

Hey,

terminate()

terminate()

is intended to terminate the instance, and although it is not supported yet in dev, it still wouldn't clear the workflow instance state.
Under

.wrangler/state/workflows

.wrangler/state/workflows

there will be 3 storage files (.sqlite, .sqllite-wal and .sqlite-shm) for each created workflow (the file names are not related to the workflow instance ids, so it is complicated to know which ones to delete). If you really want to run a workflow instance with the id equal to one that already ran, your best option is to delete the files under that directory and re-run the workflow instance.

OOlga Silva Hey, `terminate()` is intended to terminate the instance, and although it is not...

vijayaraagavan•9/15/25, 12:11 PM

Thank you!

orangesbracelet•9/15/25, 12:32 PM

hi folks, I had cronjob which takes more than 15 minutes so I refactored to workflow as there is no duration limit for steps. I was triggering workflow creation inside cronjob. But then I stuck at "too many subrequests". I send about 3k request to an api. In this case what should I do, did you have any problem like this ?

Oorangesbracelet hi folks, I had cronjob which takes more than 15 minutes so I refactored to work...

Matt Silverlock•9/15/25, 1:43 PM

Batch those requests into separate Workflow instance runs - e.g. instead of 3000 in one, fire off 3-4 workflows that each handle < 1000 requests to that API (https://developers.cloudflare.com/workflows/reference/limits/).

OOllie Is there any news on having singleton workflows / limiting a specifically workfl...

avenceslau•9/15/25, 2:51 PM

On our roadmap, stay tuned! :soontm:

Aavenceslau On our roadmap, stay tuned! :soontm:

OllieOP•9/15/25, 3:02 PM

Ah amazing, I run around 2 million workflows a month, and have more to add

, but since they use external services I need to rate limit them

Victor•9/15/25, 6:03 PM

Found a bug in making workflows. Despite the docs saying you can't, you can make workflows with ids beyond 64 characters and outside of the regex listed https://developers.cloudflare.com/workflows/reference/limits/

jordanfinners•9/16/25, 6:19 AM

Hey, I wondered if you knew if it was possible to increase the memory limit of a workflow step? Please

I've requested via the form but not heard back.

My context is I'm processing data for my customers which can range from a few KB to a 400MB json blob.
At the moment I'm using streams to get this into R2 and some processing to reduce down the total size here.
However I then need to do a load of comparisons between this and another data set. I've tried not loading into memory and streaming but then just means I hit CPU limits when using streams rather than memory

Jjordanfinners Hey, I wondered if you knew if it was possible to increase the memory limit of a...

avenceslau•9/16/25, 10:48 AM

We currently cannot increase those limits. We are working on it though.

jordanfinners•9/16/25, 10:49 AM

Thank you for the info! Do you have a rough timeline, like first half of next year etc, for when any increases will be possible? Please

xenophyte1234•9/16/25, 4:58 PM

Hey, I got errors like instance.not_found for my workflow instance when I query

env.WORKFLOW.get(instanceId)

env.WORKFLOW.get(instanceId)

, while the instance is on the dashboard. Is there anything wrong? I also found occationly the dashboard page shows my workflow instance status is unknown but display it normally after I refreshed the page.

duxum•9/17/25, 6:27 AM

Greetings,
I am calling like this to start a workflow.

const instances = await env.MATCH_DAY_UPDATE.createBatch(
  batch.map((match) => ({ params: { matchId: match.id } })),
);

const instances = await env.MATCH_DAY_UPDATE.createBatch(
  batch.map((match) => ({ params: { matchId: match.id } })),
);

But the instances value is an empty array and no workflow is starting yet batch contain some elements.

Is there a problem?

Thanks,

oscar•9/17/25, 10:47 AM

Greetings, having the same issue as some others here, where valid Workflow IDs are visible in the UI but can't be found by my Worker.

Workflow name is: PipelineV2Runner-production
A Workflow ID this happened to is: e215050f-edb0-4132-9484-9da1e84019ea

I suspected some kind of race condition and tried a workaround where failed .get operations would be sent to a Queue and retried. This managed to recover some Workflows for a few minutes but has then been completely failing too.

Olriko•9/17/25, 3:19 PM

Hello,
Getting unhelpful error messages from Cloudflare Workflows.
The logs only show generic "run" errors without any actual error details or stack traces:

{ "source": { "level": "error", "message": "run" }, "dataset": "cloudflare-workers", "timestamp": "2025-09-17T14:51:22.136Z", "$workers": { "diagnosticsChannelEvents": [], "truncated": false, "event": { "rpcMethod": "run" }, "scriptName": "worker-insight-staging", "outcome": "exception", "eventType": "rpc", "entrypoint": "FetchAdInsightsWorkflow", "scriptVersion": { "id": "f85a34a5-8a00-4692-b0bc-4c80b91fe8ef" }, "executionModel": "stateless", "requestId": "PUFWDN8JVB9SKDFI", "wallTimeMs": 463, "cpuTimeMs": 3 }, "$metadata": { "id": "01K5C2H8MR4QRA6B5HE0KF1FBX", "requestId": "PUFWDN8JVB9SKDFI", "trigger": "default.run", "service": "worker-insight-staging", "level": "error", "error": "run", "message": "run", "account": "e97e3fed722428f4d726062eae9aa766", "type": "cf-worker-event", "fingerprint": "34bb94fedcdf1b63ffea70b314aa379b", "origin": "rpc", "messageTemplate": "run" }, "links": [] }

{ "source": { "level": "error", "message": "run" }, "dataset": "cloudflare-workers", "timestamp": "2025-09-17T14:51:22.136Z", "$workers": { "diagnosticsChannelEvents": [], "truncated": false, "event": { "rpcMethod": "run" }, "scriptName": "worker-insight-staging", "outcome": "exception", "eventType": "rpc", "entrypoint": "FetchAdInsightsWorkflow", "scriptVersion": { "id": "f85a34a5-8a00-4692-b0bc-4c80b91fe8ef" }, "executionModel": "stateless", "requestId": "PUFWDN8JVB9SKDFI", "wallTimeMs": 463, "cpuTimeMs": 3 }, "$metadata": { "id": "01K5C2H8MR4QRA6B5HE0KF1FBX", "requestId": "PUFWDN8JVB9SKDFI", "trigger": "default.run", "service": "worker-insight-staging", "level": "error", "error": "run", "message": "run", "account": "e97e3fed722428f4d726062eae9aa766", "type": "cf-worker-event", "fingerprint": "34bb94fedcdf1b63ffea70b314aa379b", "origin": "rpc", "messageTemplate": "run" }, "links": [] }

This makes debugging impossible. Is there a way to get actual error messages/stack traces from Workflows? Or is this a known logging issue?

It's a log from the worker. from the workflow page, there is no errors.

OOlriko Hello, Getting unhelpful error messages from Cloudflare Workflows. The logs only...

Seekerdasbatatas•9/17/25, 3:37 PM

This is an unfortunate side effect of how we guarantee durable execution, and we are trying to improve the logging pipeline so that they are not noisy for y'all

:soontm: - my recommendation for a workaround is to filter the errors from logs that don't contain error messages or stacks

SSeekerdasbatatas This is an unfortunate side effect of how we guarantee durable execution, and we...

Olriko•9/17/25, 4:17 PM

oh ok. so this log is not an error.
It should be awesome if logs can be displayed into the step or instance view :MeowHeartCloudflare:

flof.fly•9/17/25, 6:34 PM

Can i use introspectWorkflow testing inside a runInDurableObject context?

If not, what is the right way to run e2e testing locally for workers that trigger workflows that have steps with DurableObject communications?

Fflof.fly Can i use introspectWorkflow testing inside a runInDurableObject context? If no...

Seekerdasbatatas•9/17/25, 6:46 PM

introspectWorkflow

introspectWorkflow

should be done at the start of the test (you can think of doing

introspectWorkflow

introspectWorkflow

where you mock other APIs and so on) - intercepted workflows with DOs should just work (if you want to mock/intercept the calls between the two, let me know)

flof.fly•9/18/25, 7:55 PM

Why workflows have such a big wall time? what could the reason be? Is this walletime expected?
Is it related to retries?

in my workflow i have one step that is expected to fail and go errored status, maybe that's the issue there?

Here's my workflow step configs:

const defaultConfig: WorkflowStepConfig = {
    retries: {
        limit: 0,
        delay: 0,
        backoff: 'constant',
    },
    timeout: '3 minutes',
}

const defaultConfig: WorkflowStepConfig = {
    retries: {
        limit: 0,
        delay: 0,
        backoff: 'constant',
    },
    timeout: '3 minutes',
}

TheJudas•9/18/25, 7:57 PM

Thinking of doing some video processing using workflows is it possible to run ffmpeg there?

TTheJudas Thinking of doing some video processing using workflows is it possible to run ff...

full_moon_flowers•9/18/25, 9:18 PM

I don't think Workflows (Workers) supports ffmpeg natively, but IIUC you could run ffmpeg from a Container as part of a workflow

TTheJudas Thinking of doing some video processing using workflows is it possible to run ff...

Shivek•9/19/25, 6:06 PM

Container, or just call a 3rd party server running ffmpeg.

Isak•9/20/25, 1:31 PM

Can you actually see the console.logs anywhere?

jordanfinners•9/20/25, 9:53 PM

I'm using wait for event in a for loop to trigger nested workflow. I see the error where the parent hasn't waited for the event on longer running workflows. Is it because I'm doing this outside of a step?

Murder Chicken•9/21/25, 2:22 PM

When workflows are triggered via cron vs. manually triggering them from the dashboard, is there any context difference? All of a sudden, I'm seeing workflows failing they are on steps that proxy requests through another worker but if I manually trigger the workflow, it completes just fine. Trying to understand what could be different between a cron triggered workflow and one triggered through the dashboard.

Murder Chicken•9/21/25, 4:43 PM

Okay, looks like I understand what's going on, when you trigger the workflow it seems to initiate the worker near your geographic location. When done through

cron

cron

it seems to do it from other geographic locations. I have a security rule to block all traffic to the worker that's not from the United States and it seems that, as of recently, my cron-scheduled workflows are being instantiated outside of the United States in Poland or Singapore which is causing them to be blocked by the security rules.

Is there a way to force location of execution of workflows or whitelist / exempt same account workers from security rules?

ajay | dreamlit.ai•9/21/25, 7:10 PM

We are seeing a handful of workflows occasionally be stuck in QUEUED. They never get out of this state. We have to manually restart them. Hasn't been an issue until about a week ago, now seeing this happen sporadically when we start workflows. Any idea what might be up?

orangesbracelet•9/24/25, 9:56 PM

hi guys, I wonder which way should we prefer,

Defining if and else blocks and inside if and else calling different steps
Creating just one step and defining if and elses inside step and calling functions inside if elses

do you have any idea

Oorangesbracelet hi guys, I wonder which way should we prefer, 1. Defining if and else blocks a...

avenceslau•9/25/25, 9:22 AM

As long as the expression on the if is deterministic you can do either.

Sscook I'm getting the following error for my workflows that have a sleep step for a lo...

orangesbracelet•9/25/25, 11:36 AM

is there anyone still get this error, we get too

bert•9/25/25, 11:57 AM

Hey

I deployed a new version of my workflow and I'm now running into

WorkflowInternalError: Attempt failed due to internal workflows error

WorkflowInternalError: Attempt failed due to internal workflows error

errors - at any step (even those that weren't affected by my latest deploy).

Run ID:

6b8af9a0-7ba7-4f2a-ade4-31500c2b2cfb

6b8af9a0-7ba7-4f2a-ade4-31500c2b2cfb

0ecea869-a90e-416f-a882-2a84f45d7f13

0ecea869-a90e-416f-a882-2a84f45d7f13

, or

5ba45a09-a860-4f7c-8097-197b67dbfbf0

5ba45a09-a860-4f7c-8097-197b67dbfbf0

.

Any idea what's going on?

面

面条•9/25/25, 12:16 PM

Starting from 09.23, my scheduled task triggered twice.

Ashish Rawat•9/25/25, 1:49 PM

Does anyone know how to make workflows work with opennext. In dev mode, I tried with both

wrangler dev

wrangler dev

next dev

next dev

but getting error

orangesbracelet•9/25/25, 6:40 PM

Is there anyone faces invalid instance id problem, in cloudflare docs I didnt see any rule about this

.r20•9/28/25, 4:03 AM

~~is there any way to view logs in a workflow?~~

~~never mind its in worker logs~~

CWhy•9/29/25, 6:20 AM

hi I also got "Attempt failed due to internal workflows error" for my workflow for instance 3f151892-90a4-446b-ad4a-394d3e44c110 and 8f204f7e-2032-446f-acee-fee00197b48b. anyway to debug this?

Yuan•9/29/25, 10:42 AM

Hi, i am implementing a feature that requires doing file conversion, and the file can be > 30mb. a single worker definetely can't handle it. haven't thought everything through, but in general i think the entire process can be divided into multiple steps (e.g. read, convert, store ...) is workflow a good solution for this kind of use case? otherwise can i have some suggestion. thanks

.r20•9/29/25, 3:32 PM

Is there any way to debug workflows?

.r20•9/29/25, 3:33 PM

it seems like onl;y the invocation islogged in workers logs

ByteBeast•9/30/25, 5:54 AM

Hello everyone, I want to know how much time duration workflow can stay running? like if I set 5 delay for one year will it stay up for 5 years?

BByteBeast Hello everyone, I want to know how much time duration workflow can stay running?...

avenceslau•9/30/25, 1:50 PM

It depends on what you mean by stay up. If you do 5 one year sleeps your workflow instance will "hibernate" for one year 5 times, yielding concurrency to other instances while it sleeps. It will awaken after every year and go back to sleep. The workflow does not use compute while it sleeps.

Frank•9/30/25, 5:31 PM

currently using a DO for threadlock so only 1 workflow is triggered per user. is there a better way?

.r20•9/30/25, 6:46 PM

why not use d1?

.r20•9/30/25, 6:47 PM

that way you dont deal with the costs associated by provisioning a DO

.r20•9/30/25, 6:48 PM

just keep some minimal d1 table

{ "source": { "level": "error", "message": "run" }, "dataset": "cloudflare-workers", "timestamp": "2025-09-17T14:51:22.136Z", "$workers": { "diagnosticsChannelEvents": [], "truncated": false, "event": { "rpcMethod": "run" }, "scriptName": "worker-insight-staging", "outcome": "exception", "eventType": "rpc", "entrypoint": "FetchAdInsightsWorkflow", "scriptVersion": { "id": "f85a34a5-8a00-4692-b0bc-4c80b91fe8ef" }, "executionModel": "stateless", "requestId": "PUFWDN8JVB9SKDFI", "wallTimeMs": 463, "cpuTimeMs": 3 }, "$metadata": { "id": "01K5C2H8MR4QRA6B5HE0KF1FBX", "requestId": "PUFWDN8JVB9SKDFI", "trigger": "default.run", "service": "worker-insight-staging", "level": "error", "error": "run", "message": "run", "account": "e97e3fed722428f4d726062eae9aa766", "type": "cf-worker-event", "fingerprint": "34bb94fedcdf1b63ffea70b314aa379b", "origin": "rpc", "messageTemplate": "run" }, "links": [] }

Is there any news on having singleton workflows / limiting a specifically workflow to concurrency of

Similar Threads