hey guys how do I reduce my storage metric in workflows? I see that I have over 5GB of storage used

hey guys how do I reduce my storage metric in workflows? I see that I have over 5GB of storage used (!!) and I do not need this but I don't see any option to delete old instances or even configure the workflow API to set a smaller retention period. Any ideas?

ccarse•4/10/25, 3:12 AM

So I'm building a tool that runs a workflow when a file gets placed in an r2 bucket. All working well. I'm trying to build a dashboard to show the upload events, trigger retries, etc.. The problem I'm running into is that env.WORKFLOW.get(<instanceId>) doesn't work for me in dev. Is this intended?

JJulianD hey guys how do I reduce my storage metric in workflows? I see that I have over ...

Matt Silverlock•4/10/25, 5:25 AM

This will be landing soon, but we don’t yet charge for storage.

Cccarse So I'm building a tool that runs a workflow when a file gets placed in an r2 buc...

Matt Silverlock•4/10/25, 5:26 AM

Can you show the code, and what does “doesn’t work” mean - what is the error message? What version of wrangler?

MMatt Silverlock Can you show the code, and what does “doesn’t work” mean - what is the error mes...

ccarse•4/10/25, 5:27 AM

I just updated wrangler to latest. The code snippet I posted just returns null when running locally.

Cccarse I just updated wrangler to latest. The code snippet I posted just returns null w...

Matt Silverlock•4/10/25, 5:28 AM

What code snippet?

Matt Silverlock•4/10/25, 5:29 AM

Post the actual code - a minimal reproduction.

MMatt Silverlock Post the actual code - a minimal reproduction.

ccarse•4/10/25, 5:29 AM

thanks for the reply. Will tomorrow. Just got in bed.

Original message was deleted

Suman!•4/10/25, 11:14 AM

Just got it working!! Awesome!! Thanks

MMatt Silverlock Post the actual code - a minimal reproduction.

ccarse•4/10/25, 2:42 PM

So after further investigation, it turns out any time the dev server reloads (any code change), every workflow instance disappears. This is really annoying. Is this expected?

You can repro with the workflows starter template easily.

npm create cloudflare@latest workflows-starter -- --template "cloudflare/workflows-starter"

npm create cloudflare@latest workflows-starter -- --template "cloudflare/workflows-starter"

Just run it, go to localhost:8787, record the instanceId and note that localhost:8787?instanceId=<id> shows details of the instance, add a console.log call anywhere and save, now localhost:8787?instanceId=<id> gives

instance.not_found

instance.not_found

error.

Aster•4/10/25, 3:48 PM

Hello, do you have a cron mode for workflows? Let's say; start workflow A every 1st monday of the month?

fredviken•4/10/25, 3:50 PM

Hi! Has anyone been able to get cross-script to work locally? I've seen other people encounter this issue but not seen any solutions

✘ [ERROR] Worker "workflows:s3-update-status"'s binding "USER_WORKFLOW" refers to a service "core:user:integrations", but no such service is defined.

✘ [ERROR] Worker "workflows:s3-update-status"'s binding "USER_WORKFLOW" refers to a service "core:user:integrations", but no such service is defined.

It works when called from the same worker, and service bindings between the workers seem to work fine, it's only an issue with the cross-script bindings.

Victor•4/10/25, 4:16 PM

When workflows are queued, is it normal behavior to not show the

params

params

Cccarse So after further investigation, it turns out any time the dev server reloads (an...

Matt Silverlock•4/10/25, 4:39 PM

That's normal: dev sessions don't persist by default (otherwise we'd be told we're filling up everyone's disk + clobbering test runs/local dev). What are you trying to solve across

wrangler dev

wrangler dev

invocations?

Ffredviken Hi! Has anyone been able to get cross-script to work locally? I've seen other pe...

Matt Silverlock•4/10/25, 4:39 PM

Show your

wrangler

wrangler

config and code.

MMatt Silverlock That's normal: dev sessions don't persist by default (otherwise we'd be told we'...

ccarse•4/10/25, 4:55 PM

I mean, I think there's a big gap between never letting things expire and expiring everything for a single character code change anywhere in the project right?

And let me be clear, this is a single

wrangler dev

wrangler dev

invocation.

Like one option is that you could have a TTL on workflow instances that defaulted to like a day or something?

Cccarse I mean, I think there's a big gap between never letting things expire and expiri...

Matt Silverlock•4/10/25, 4:57 PM

I can't reproduce that in the same invocation (nor should I be able to): instances persist during the running

wrangler dev

wrangler dev

run:

Matt Silverlock•4/10/25, 4:57 PM

If you change the code, that's a new invocation (reload of all state)

ccarse•4/10/25, 4:58 PM

Oh I wouldn't call that a new invocation, I only invoked it once.

Matt Silverlock•4/10/25, 4:58 PM

I understand where you're coming from, but I'm also explaining why this is the case

ccarse•4/10/25, 4:58 PM

Like, for example, my kv data doesn't disappear on each invocation right?

Matt Silverlock•4/10/25, 4:58 PM

Will chat w/ the team about options to opt-in to persistency here

ccarse•4/10/25, 4:58 PM

Why do instances?

Cccarse Why do instances?

Matt Silverlock•4/10/25, 4:59 PM

Instances aren't just state on disk (unlike KV pairs) - again, understand the frustration but not the same thing.

ccarse•4/10/25, 5:02 PM

I really like the products you all are putting out but it's frustrating how every product behaves differently in terms of local vs remote vs what is default and what's available. I know it's early days for workflows but just wanted to give the feedback.

ccarse•4/10/25, 5:29 PM

Is there a way to list recently invoked workflow instances in dev? I can use the api in prod but that doesn't work for dev.

MMatt Silverlock Show your `wrangler` config and code.

fredviken•4/10/25, 5:47 PM

These are the wrangler configs. the workflow is just a simple placeholder one for now with a console log and sleep, but it works like expected when called in the same worker. But the second i add the binding in the other worker it crashes, despite being able to connect using service bindings

Ffredviken These are the wrangler configs. the workflow is just a simple placeholder one fo...

Matt Silverlock•4/10/25, 6:17 PM

Thanks - taking a look!

MMatt Silverlock Call `.create` as needed for each Workflow.

hugo martins•4/10/25, 6:58 PM

Will this make only one request? If not, then there's a possibility that one of the requests might fail.

Victor•4/10/25, 7:03 PM

I'm trying to list or terminate workflows (both via rest api and via dashboard) returns

HTTP 500/10001 workflows.api.error.internal_server

HTTP 500/10001 workflows.api.error.internal_server

Victor•4/10/25, 7:08 PM

Now getting 503 service unavailable (7010 service unavailable)

Victor•4/10/25, 7:15 PM

Ok, any workflows action, even creating workflows from inside a worker (via binding) throws errors

Hhugo martins Will this make only one request? If not, then there's a possibility that one of ...

Matt Silverlock•4/10/25, 7:36 PM

One request per Workflow. You can retry any that fail (although that should be very rare). How would you expect this to be solved?

MMatt Silverlock @Workflows - Workflows is GA + we've shipped the new `step.waitForEvent` (and `i...

ajay | dreamlit.ai•4/10/25, 9:53 PM

Wooo! Congrats on the launch excited for this

Ffredviken These are the wrangler configs. the workflow is just a simple placeholder one fo...

felixnorden•4/10/25, 10:01 PM

Getting similar issues when using the Vitest Pool Workers lib for testing my workflow that's triggered by a Durable Object.

felixnorden•4/10/25, 10:01 PM

Lmk if you solve this, it has me stuck :kek:

MMatt Silverlock @Workflows - Workflows is GA + we've shipped the new `step.waitForEvent` (and `i...

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/10/25, 10:02 PM

env

env

inside a workflow context? we would like to use our KV bindings

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/10/25, 10:03 PM

congrats on the launch btw!

Ss̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜is `env` inside a workflow context? we would like to use our KV bindings

Matt Silverlock•4/10/25, 10:04 PM

this.env

this.env

ensuring you pass your
```
Env
```
```
Env
```
type as a parameter to your Workflow class: https://developers.cloudflare.com/workflows/build/workers-api/#workflowentrypoint

Cloudflare Docs

Workers API · Cloudflare Workflows docs

This guide details the Workflows API within Cloudflare Workers, including methods, types, and usage examples.

Matt Silverlock•4/10/25, 10:04 PM

Example usage: https://developers.cloudflare.com/workflows/examples/backup-d1/

Cloudflare Docs

Export and save D1 database · Cloudflare Workflows docs

Send invoice when shopping cart is checked out and paid for

MMatt Silverlock `this.env` - ensuring you pass your `Env` type as a parameter to your Workflow c...

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/10/25, 10:05 PM

good to know, thanks! don't think i saw this documented explicitly btw

Ss̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜good to know, thanks! don't think i saw this documented explicitly btw

Matt Silverlock•4/10/25, 10:05 PM

Let me fix!

MMatt Silverlock Thanks - taking a look!

felixnorden•4/10/25, 10:47 PM

Do you happen to know how to get passed this issue with running tests with the Vitest Pool Workers library with Workflows that are bound to a Worker and DO? Adding it in causes a similar error to what fredviken is getting. Do I need to do any extra imports in the

vitest.config

vitest.config

file as I have the workflow in a separate folder? The durable object class gets resolved without issue, but not the workflow.

Snippet of my wrangler.jsonc below

{
  // Other info
  // ...
  "compatibility_date": "2025-03-17",
  "compatibility_flags": ["nodejs_compat"],

  "observability": {
    "enabled": true
  },

  "durable_objects": {
    "bindings": [
      {
        "name": "TASKS",
        "class_name": "Task"
      }
    ]
  },

  "migrations": [
    {
      "new_sqlite_classes": ["Task"],
      "tag": "v1"
    }
  ],

  "workflows": [
    {
      "binding": "TASK_WORKFLOW",
      "class_name": "TaskWorkflow",
      "name": "task-workflow"
    }
  ],
// Some other bindings with Queues, Services and Rules
// ...
}

{
  // Other info
  // ...
  "compatibility_date": "2025-03-17",
  "compatibility_flags": ["nodejs_compat"],

  "observability": {
    "enabled": true
  },

  "durable_objects": {
    "bindings": [
      {
        "name": "TASKS",
        "class_name": "Task"
      }
    ]
  },

  "migrations": [
    {
      "new_sqlite_classes": ["Task"],
      "tag": "v1"
    }
  ],

  "workflows": [
    {
      "binding": "TASK_WORKFLOW",
      "class_name": "TaskWorkflow",
      "name": "task-workflow"
    }
  ],
// Some other bindings with Queues, Services and Rules
// ...
}

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/10/25, 11:38 PM

Are there guarantees that in flight workflow requests are allowed to finish when a new worker version is deployed?

We're ready to migrate off Durable Objects in favor of Workflows because DOs don't have this guarantee; just wanted to make sure that Workflows actually do have this property

ddsgadget•4/11/25, 2:02 AM

waitForEvent seems really powerful, but I can't figure out how it works exactly. To trigger the event, you need to know its instanceId, await env.MY_WORKFLOW.get(instanceId)m , but how will the external service know the instanceId that is determined by the workflow internally? Your example with Stripe here: https://developers.cloudflare.com/workflows/build/events-and-parameters/#send-events-to-running-workflows is interesting, but how would Stripe know what the instance id is to trigger?

Cloudflare Docs

Events and parameters · Cloudflare Workflows docs

When a Workflow is triggered, it can receive an optional event. This event can include data that your Workflow can act on, including request details, user data fetched from your database (such as D1 or KV) or from a webhook, or messages from a Queue consumer.

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/11/25, 3:50 AM

Can multiple workers bind to workflows sharing the same name? For preview deploys, we currently spin up a new worker, but it shares a wrangler.toml as our production deploy, so we would have multiple workers with workflows of the same name.

This UI seems to imply that a given workflow can only have a single bound worker?

EDIT: seems like it can't. templating in our wrangler.toml seems to fix the issue...

inaridiy•4/11/25, 5:49 AM

Are there any plans to lift or increase the current limit of 1,000 external requests per workflow instance? Right now, I'm having to split my workflows just to work around this restriction.

s̸̘̩̓̍ẻ̸͕ͅå̶̡͖ṅ̴̹̊͜•4/11/25, 6:22 AM

Bug report: workflows are not created in cloudflare when using gradual deploys, at least when created for the first time

Dddsgadget waitForEvent seems really powerful, but I can't figure out how it works exactly....

Matt Silverlock•4/11/25, 6:24 AM

In most cases you likely want a Worker to do the routing from webhook -> the correct instance, but if (for example) you can attach metadata to the Stripe payment event or subscription request, you can include the instance ID.

Iinaridiy Are there any plans to lift or increase the current limit of 1,000 external requ...

Matt Silverlock•4/11/25, 6:25 AM

Yes, but will take a bit of time (weeks, not days).

Avi•4/11/25, 7:53 AM

now that workflows are GA/production ready, how can i delete a workflow?

{ // Other info // ... "compatibility_date": "2025-03-17", "compatibility_flags": ["nodejs_compat"], "observability": { "enabled": true }, "durable_objects": { "bindings": [ { "name": "TASKS", "class_name": "Task" } ] }, "migrations": [ { "new_sqlite_classes": ["Task"], "tag": "v1" } ], "workflows": [ { "binding": "TASK_WORKFLOW", "class_name": "TaskWorkflow", "name": "task-workflow" } ], // Some other bindings with Queues, Services and Rules // ... }

hey guys how do I reduce my storage metric in workflows? I see that I have over 5GB of storage used

Similar Threads