Is it possible to get the number of times a workflow step has retried? I'd like to call some code on

K

koraytaylan•5/24/25, 3:39 PM

in workflow instance details page, output gets truncated for a particular "step history" which is a problem for me. i wonder if there is any plans to change this behavior?

M

Mr. Technician•5/24/25, 4:35 PM

I have a question the docs are not clear on:

// ✅ Good: steps that are dynamically named are constructed in a deterministic way.
    // In this case, `catList` is a step output, which is stable, and `catList` is
    // traversed in a deterministic fashion (no shuffles or random accesses) so,
    // it's fine to dynamically name steps (e.g: create a step per list entry).
    let catList = await step.do("get cat list from KV", async () => {
      return await env.KV.get("cat-list");
    });

    for (const cat of catList) {
      await step.do(`get cat: ${cat}`, async () => {
        return await env.KV.get(cat);
      });
    }

// ✅ Good: steps that are dynamically named are constructed in a deterministic way.
    // In this case, `catList` is a step output, which is stable, and `catList` is
    // traversed in a deterministic fashion (no shuffles or random accesses) so,
    // it's fine to dynamically name steps (e.g: create a step per list entry).
    let catList = await step.do("get cat list from KV", async () => {
      return await env.KV.get("cat-list");
    });

    for (const cat of catList) {
      await step.do(`get cat: ${cat}`, async () => {
        return await env.KV.get(cat);
      });
    }

// ✅ Good: steps that are dynamically named are constructed in a deterministic way.
    // In this case, `catList` is a step output, which is stable, and `catList` is
    // traversed in a deterministic fashion (no shuffles or random accesses) so,
    // it's fine to dynamically name steps (e.g: create a step per list entry).
    let catList = await step.do("get cat list from KV", async () => {
      return await env.KV.get("cat-list");
    });

    for (const cat of catList) {
      await step.do(`get cat: ${cat}`, async () => {
        return await env.KV.get(cat);
      });
    }

// ✅ Good: steps that are dynamically named are constructed in a deterministic way.
    // In this case, `catList` is a step output, which is stable, and `catList` is
    // traversed in a deterministic fashion (no shuffles or random accesses) so,
    // it's fine to dynamically name steps (e.g: create a step per list entry).
    let catList = await step.do("get cat list from KV", async () => {
      return await env.KV.get("cat-list");
    });

    for (const cat of catList) {
      await step.do(`get cat: ${cat}`, async () => {
        return await env.KV.get(cat);
      });
    }

Can I store the looped returned values? If so, how can I do so deterministically?

MMr. Technician I have a question the docs are not clear on: ```ts // ✅ Good: steps that are dy...

K

koraytaylan•5/24/25, 5:34 PM

in the workflow rules page there was an example to that. the correct way was to use

Promise.all

Promise.all

Promise.all

Promise.allif i recall correctly.

Kkoraytaylan in the workflow rules page there was an example to that. the correct way was to ...

M

Mr. Technician•5/24/25, 5:35 PM

that example is different though, it used a known number of steps (2 in the example)

M

Mr. Technician•5/24/25, 5:35 PM

unless I can just slap this loop into a promise.all?

M

Mr. Technician•5/24/25, 5:35 PM

I'm more experienced with Tasks in C# than promises

MMr. Technician that example is different though, it used a known number of steps (2 in the exam...

K

koraytaylan•5/24/25, 5:46 PM

ok i see your point. to store the returned items you need to do,

const cats = await Promise.all(catList.map(cat => step.do(`get cat: ${cat}`, () => env.KV.get(cat))));

const cats = await Promise.all(catList.map(cat => step.do(`get cat: ${cat}`, () => env.KV.get(cat))));

const cats = await Promise.all(catList.map(cat => step.do(`get cat: ${cat}`, () => env.KV.get(cat))));

const cats = await Promise.all(catList.map(cat => step.do(`get cat: ${cat}`, () => env.KV.get(cat))));

this should work

Kkoraytaylan ok i see your point. to store the returned items you need to do, ```const cats =...

M

Mr. Technician•5/24/25, 8:14 PM

I'll give it a try, thanks

M

Marco Aleixo•5/24/25, 11:53 PM

I'm facing a lot of

A call to blockConcurrencyWhile() in a Durable Object waited for too long. The call was canceled and the Durable Object was reset.

while seeing spikes on my file upload endpoint, and this is happening on something internal on my Workflows, and those errors seems to be not retriable, how fix this?

A

acOP•5/26/25, 2:03 PM

I have some logic that returns from a Workflow

run

run

run

run function after 2 steps if it doesn't need to progress further -- not an error, just a situation where it doesn't need to continue. I'm not throwing an error, just returning an object with some data, but for some reason it seems like this is still showing in the workflow's worker logs as an "Error" (but it shows up correctly in the Workflow logs as "Completed" -- no error). Is there something specific I need to be returning from the

run

run

run

run function so it doesn't register as an error/Exception in the Worker?

Aac I have some logic that returns from a Workflow `run` function after 2 steps if i...

C

Chaika•5/26/25, 2:08 PM

It's just how they all show, team was working on a fix
workflows
https://discord.com/channels/595317990191398933/1303411588929294358/1303488051917492318

CChaika It's just how they all show, team was working on a fix https://discord.com/chann...

A

acOP•5/26/25, 2:09 PM

Ah thank you, good to know. That message being 6 months old doesn't give me a ton of confidence

Aac Is it possible to get the number of times a workflow step has retried? I'd like ...

J

johtso•5/27/25, 12:13 AM

You can catch the error (meaning retries were exhausted), do whatever you need to do, then rethrow the error

A

alex (he/him)•5/27/25, 7:02 AM

Can I use a Tail Worker to handle logs from workflow steps?

A

alex (he/him)•5/27/25, 7:02 AM

It seems that the logs don't get send to the tail worker.

Aalex (he/him)It seems that the logs don't get send to the tail worker.

A

acOP•5/27/25, 2:24 PM

The logs should be in the main worker that the workflow is attached to

D

ddsgadget•5/28/25, 7:29 AM

@alex (he/him) I have looked everywhere and can never find logs for any steps in a workflow. Maybe they are somewhere but I cant find them. This is single biggest weakness of workflows right now - logs are incredibly difficult to find for steps and this makes debugging them close to impossible. they should put the logs for each step on the actual step in the dashboard. This is what most other workflow platforms do - the logs for any workflow step are immediately visible in a UI for a step. i guess the only thing to do is send the logs to an outside service from each step, which make things complicated.

P

Paul Mihai•5/28/25, 9:54 AM

This may be a bit off topic but I have a question for CF techs - why don't you guys buy/partner with the n8n project and integrate it over your services? Technologically it looks like a marriage made in heaven! It's almost like they built it for cloudlfare!

K

kacper•5/28/25, 10:09 AM

Hey, I'm implementing some workflow

await step.do("process-events", async () => {
  const eventsGroupedBySomething = [[], [], [], [], [], []]; // potentially ~10 groups with 100s of events each

  for (const events of eventsGroupedBySomething) {
    for (let i = 0; i < events.length; i++) {
      const event = events[i]!;
      const { embedding, summary } = await generateEventSummary(event);
      await db
        .insert(schema.eventVector)
        .values({ id: nanoid(), eventId: event.id, embedding, summary });
    }
  }
});

await step.do("process-events", async () => {
  const eventsGroupedBySomething = [[], [], [], [], [], []]; // potentially ~10 groups with 100s of events each

  for (const events of eventsGroupedBySomething) {
    for (let i = 0; i < events.length; i++) {
      const event = events[i]!;
      const { embedding, summary } = await generateEventSummary(event);
      await db
        .insert(schema.eventVector)
        .values({ id: nanoid(), eventId: event.id, embedding, summary });
    }
  }
});

await step.do("process-events", async () => {
  const eventsGroupedBySomething = [[], [], [], [], [], []]; // potentially ~10 groups with 100s of events each

  for (const events of eventsGroupedBySomething) {
    for (let i = 0; i < events.length; i++) {
      const event = events[i]!;
      const { embedding, summary } = await generateEventSummary(event);
      await db
        .insert(schema.eventVector)
        .values({ id: nanoid(), eventId: event.id, embedding, summary });
    }
  }
});

await step.do("process-events", async () => {
  const eventsGroupedBySomething = [[], [], [], [], [], []]; // potentially ~10 groups with 100s of events each

  for (const events of eventsGroupedBySomething) {
    for (let i = 0; i < events.length; i++) {
      const event = events[i]!;
      const { embedding, summary } = await generateEventSummary(event);
      await db
        .insert(schema.eventVector)
        .values({ id: nanoid(), eventId: event.id, embedding, summary });
    }
  }
});

and I'd want to wrap the inner loop in a

step.do

step.do

step.do

step.do for retries but then I could really quickly hit 1024 steps limit. Maybe I could invoke separate workflow in the inner loop to just handle processing the event? But then my

generateEventSummary

generateEventSummary

generateEventSummary

generateEventSummary might throw 429 error. I guess sending queue message for every event and handling it in separate worker might be a better approach? What would be the "idiomatic" CF approach? Thanks!

F

Fine Euro•5/28/25, 10:29 AM

Hello everyone I'm alex,

B

Ben•5/29/25, 5:41 PM

Hello, we’re trying to terminate a workflow and it’s not working as expected so I wanted to check before we investigate further on our side - is the

.terminate()

.terminate()

.terminate()

.terminate() method on a workflow instance implemented at the moment?

J

jonast•5/29/25, 8:03 PM

Has anyone seen transient error 401 Unauthorised when trying to create workflows via the cloudflare API?

I'm using Alchemy and about half the time I get this when attempting to create my workflow resources

Scope failed estate/banana--prod/worker/worker
Scope is in error, skipping finalize
Scope failed estate/banana--prod/worker
Scope failed estate/banana--prod/worker
Scope is in error, skipping finalize
17 |     errorData;
18 |     /**
19 |      * Create a new CloudflareApiError
20 |      */
21 |     constructor(message, response, errorData) {
22 |         super(message);
             ^
CloudflareApiError: Error 401 create workflow 'banana--prod--GENERIC_BROWSER_WORKFLOW': Authentication error
     status: 401,
 statusText: "Unauthorized",
  errorData: [
  [Object ...]
],
      at new CloudflareApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:22:9)
      at handleApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:53:11)

Bun v1.2.15 (Linux x64 baseline)
 ELIFECYCLE  Command failed with exit code 1.

Scope failed estate/banana--prod/worker/worker
Scope is in error, skipping finalize
Scope failed estate/banana--prod/worker
Scope failed estate/banana--prod/worker
Scope is in error, skipping finalize
17 |     errorData;
18 |     /**
19 |      * Create a new CloudflareApiError
20 |      */
21 |     constructor(message, response, errorData) {
22 |         super(message);
             ^
CloudflareApiError: Error 401 create workflow 'banana--prod--GENERIC_BROWSER_WORKFLOW': Authentication error
     status: 401,
 statusText: "Unauthorized",
  errorData: [
  [Object ...]
],
      at new CloudflareApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:22:9)
      at handleApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:53:11)

Bun v1.2.15 (Linux x64 baseline)
 ELIFECYCLE  Command failed with exit code 1.

Scope failed estate/banana--prod/worker/worker
Scope is in error, skipping finalize
Scope failed estate/banana--prod/worker
Scope failed estate/banana--prod/worker
Scope is in error, skipping finalize
17 |     errorData;
18 |     /**
19 |      * Create a new CloudflareApiError
20 |      */
21 |     constructor(message, response, errorData) {
22 |         super(message);
             ^
CloudflareApiError: Error 401 create workflow 'banana--prod--GENERIC_BROWSER_WORKFLOW': Authentication error
     status: 401,
 statusText: "Unauthorized",
  errorData: [
  [Object ...]
],
      at new CloudflareApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:22:9)
      at handleApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:53:11)

Bun v1.2.15 (Linux x64 baseline)
 ELIFECYCLE  Command failed with exit code 1.

Scope failed estate/banana--prod/worker/worker
Scope is in error, skipping finalize
Scope failed estate/banana--prod/worker
Scope failed estate/banana--prod/worker
Scope is in error, skipping finalize
17 |     errorData;
18 |     /**
19 |      * Create a new CloudflareApiError
20 |      */
21 |     constructor(message, response, errorData) {
22 |         super(message);
             ^
CloudflareApiError: Error 401 create workflow 'banana--prod--GENERIC_BROWSER_WORKFLOW': Authentication error
     status: 401,
 statusText: "Unauthorized",
  errorData: [
  [Object ...]
],
      at new CloudflareApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:22:9)
      at handleApiError (/home/runner/work/monorepo/monorepo/node_modules/.pnpm/alchemy@0.20.1_@ai-sdk+openai-compatible@0.2.13_zod@3.25.17__@ai-sdk+openai@1.1.5_zod@3_72bf5a10489ada0d229ba0333e766a7c/node_modules/alchemy/lib/cloudflare/api-error.js:53:11)

Bun v1.2.15 (Linux x64 baseline)
 ELIFECYCLE  Command failed with exit code 1.

SSeekerdasbatatas hey everyone, I've done on my free time a **batteries-included** library for Wor...

A

acOP•5/30/25, 5:58 PM

Hi Luis - are you still working on flowcesinha (or any word on when any of these things will get integrated into Workflows)?

Aac Hi Luis - are you still working on flowcesinha (or any word on when any of these...

S

Seekerdasbatatas•5/30/25, 6:01 PM

I'm going to admit that I'm not having the cycles that I would like to work on it - but if there's something broken please let me know :MeowHeartCloudflare: - if you have any ideas about possible features do let me know too, about integration idk haha

SSeekerdasbatatas I'm going to admit that I'm not having the cycles that I would like to work on i...

A

acOP•5/30/25, 6:02 PM

Ok! The main thing I was looking at was the Sentry stuff, that's how I stumbled on it. Having a tough time getting Sentry to work properly within Workflows

A

acOP•5/30/25, 6:03 PM

But not an issue with your lib, just wanted to check before trying it

J

Justin•5/30/25, 6:07 PM

Anyone else having issues deploying a workflows worker? Even when logged into wrangler with a user with Super Admin permissions I'm getting an error "A request to the Cloudflare API (/accounts/XXXX/workflows/XXXX) failed." I've changed nothing in my wrangler.jsonc since the last successful deploy. Other (non-workflow) workers are deploying fine.

7

7777777•5/30/25, 6:08 PM

Maybe I missed a step, but do I need to manually setup a Service binding for my workflow? I have a worker which accepts fetch requests, and the same worker defines 3 different workflows. Locally everything seems to run fine via the toml file. When I deploy I can see the workflow deploys happening, however when I goto

<MY_WORKER>/production/settings#bindings

<MY_WORKER>/production/settings#bindings

then workflows are not shown. I was going to add them manually as a service, but am I going to have to do this for every deployment?

Kkacper Hey, I'm implementing some workflow ```typescript await step.do("process-events...

7

7777777•5/30/25, 6:12 PM

you could use kv to store state and track which

events

events

and

have been inserted already, so that way on retry you would just check the kv for where to start. - depending on db you might also want to do all the inserts as a batch after

events.length

events.length

has been processed, which would reduce the number of kv writes to track

.

JJustin Anyone else having issues deploying a workflows worker? Even when logged into wr...

T

Tait•5/30/25, 6:13 PM

Yah we have a discussion going about this in the #wrangler channel

T

Tait•5/30/25, 6:13 PM

seems to just happen with workflow deployments (not regular workers)

TTait Yah we have a discussion going about this in the #wrangler channel

J

Justin•5/30/25, 6:16 PM

Thanks! Will follow along there..

M

mochachai•5/30/25, 6:17 PM

I have a lot of weird stuff going on. My prod workflow is suddenly not bound anymore to prod DB but the binding seems to point at the dev DB? really weird

T

Tait•5/30/25, 6:17 PM

We I say "we", no one from cloudflare has weighed in just yet. Hoping to hear something soon. Not sure if/how to escalate

JJustin Anyone else having issues deploying a workflows worker? Even when logged into wr...

A

acOP•5/30/25, 6:19 PM

Yes, I'm having the same issue right now. Must be a Cloudflare problem

J

James•5/30/25, 6:44 PM

I've escalated, thanks folks. Hopefully we'll see a resolution shortly.

W

Walshy•5/30/25, 6:51 PM

https://www.cloudflarestatus.com/incidents/6xdvg9z3qnjy

Issues Deploying Workflows

W

Walshy•5/30/25, 7:47 PM

You should be able to deploy now :)

Sorry for the issue!

A

acOP•5/30/25, 7:54 PM

Looks like it's working for me now

7

7777777•5/31/25, 12:18 AM

when I call

await env.MY_WORKFLOW.createBatch(listOfInstances);

await env.MY_WORKFLOW.createBatch(listOfInstances);

it's returning

[{"fetcher":{}}]

[{"fetcher":{}}]

instead of WorkflowInstances?

7

7777777•5/31/25, 1:38 AM

The workflow status

errored

errored

is both a terminal and transient state. There should be a separate state signalling the workflow has exhausted all retries.

Hharisvsulaiman I lost all my bindings to the workfkow

M

Matt Silverlock•5/31/25, 10:43 AM

Can you give us more info? When you deploy, what happens?

Show the output of wrangler - does it pick up the bindings?

Is it possible to get the number of times a workflow step has retried? I'd like to call some code on

Similar Threads

Similar Threads

Similar Threads