nevermind, i needed to check for list_complete before, otherwise its not alwaysreturning the cursor

CCelestial Rose nevermind, i needed to check for list_complete before, otherwise its not alwaysr...

T

thomasgauvin•11/6/25, 3:48 PM

Thanks for letting us know, let us know if you run into anything else!

P

PhoenixSpace•11/6/25, 3:56 PM

Are there more KV upper tiers on the business plan? I'm storing a lot of per-user configuration data in KV (both read and write), which means occasionally getting cold reads since its locally cached but can be modified in other places. Response times in APAC and OC (Australia especially) are rough. I had some friends in Japan and Australia hit a benchmarking endpoint (i know timing across I/O isn't the most precise, but it seems to be relatively accurate given the endpoint resolution time)

300-400ms reads and 1500ms writes aren't exactly the end of the world, especially with waitUntil, but it is frustrating for some users. Thought I'd ask before looking into migrating to per-user durable objects

PPhoenixSpace Are there more KV upper tiers on the business plan? I'm storing a lot of per-use...

1

1984 Ford Laser•11/6/25, 9:07 PM

Are those read times done immediately after writing? Keen in mind KV takes 60 seconds to be globally replicated

11984 Ford Laser Are those read times done immediately after writing? Keen in mind KV takes 60 se...

P

PhoenixSpace•11/6/25, 9:13 PM

Yeah the eventual consistency isn’t an issue. The problem is more per user data could be updated overnight, and when they check during the day, it hits a cold read

P

PhoenixSpace•11/6/25, 9:13 PM

Read your own write is extraordinarily performant. It’s just cold reads

1

1984 Ford Laser•11/6/25, 9:19 PM

Only way to avoid cold reads is to regularly read I guess

DOs would be a good use case as long as the instances are spawning close to users (I've had this issue recently where I had DOs spawn on the other side of Australia to me, was seeing 100s of ms extra request time)

T

thipperz•11/7/25, 6:29 PM

Hi everyone, I'm using KV to cache some heavy db calls in my app, I have something like

const cached = await KV.get(key, { type: 'json', cacheTtl: 1800 })
if (cached) return cached

await KV.put(key, JSON.stringify(value), { expirationTtl: 1800 })

const cached = await KV.get(key, { type: 'json', cacheTtl: 1800 })
if (cached) return cached

await KV.put(key, JSON.stringify(value), { expirationTtl: 1800 })

const cached = await KV.get(key, { type: 'json', cacheTtl: 1800 })
if (cached) return cached

await KV.put(key, JSON.stringify(value), { expirationTtl: 1800 })

const cached = await KV.get(key, { type: 'json', cacheTtl: 1800 })
if (cached) return cached

await KV.put(key, JSON.stringify(value), { expirationTtl: 1800 })

Even with both

cacheTtl

cacheTtl

cacheTtl

cacheTtl and

expirationTtl

expirationTtl

expirationTtl

expirationTtl set to 1800 (30 min), the D1 analytics show hundreds of reads over 48 hours, even though the value rarely changes and should be served from cache most of the time. For example I expected around 96 calls but i'm getting like ~500.

I think I might be misunderstanding something about how KV works, is there some gotchas I must be aware of? Thanks

Tthipperz Hi everyone, I'm using KV to cache some heavy db calls in my app, I have somethi...

T

thomasgauvin•11/10/25, 5:39 PM

KV does store the key value pairs for as long as the expiration ttl, and it should return the values so we'd probably need more of your code to understand why you're going to D1 instead of KV (because it should be in KV)

Tthomasgauvin KV does store the key value pairs for as long as the expiration ttl, and it shou...

T

thipperz•11/12/25, 2:48 AM

Got it, I'm testing to see if the issue isn't somewhere else before I ask for further help

T

thipperz•11/12/25, 2:52 AM

One other thing: i tried to delete a KV Namespace from the Dashboard, but get a error message saying

remove namespace: 'namespace has associated scripts: pages-worker--2406846-production

remove namespace: 'namespace has associated scripts: pages-worker--2406846-production

remove namespace: 'namespace has associated scripts: pages-worker--2406846-production

remove namespace: 'namespace has associated scripts: pages-worker--2406846-production

However I don't have any pages project anymore (I used to have one before I moved to open-next with Workers instead). I checked and none of my current Workers use that namespace anymore.

C

chesha1•11/17/25, 3:08 AM

Bulk update can only be used with the REST API or the Wrangler CLI for now, which means I need to create an API key. I’m wondering if this can be used with Workers KV bindings in the future, as that would be more convenient.

D

ddsgadget•11/18/25, 11:45 AM

Can I assume KV is done again? All my workers with KV are failing with KV errors, and can't log into the Cloudflare Dashboard either, which is exactly what happened last time. I believe the dashboard is powered by some KV in the background.

A

AZipr•11/18/25, 12:36 PM

Likewise. It resumes for a moment then disappears

T

thomasgauvin•11/20/25, 3:35 PM

This was related to the outage on November 18, here is the report: https://blog.cloudflare.com/18-november-2025-outage/

The Cloudflare Blog

Cloudflare outage on November 18, 2025

Cloudflare suffered a service outage on November 18, 2025. The outage was triggered by a bug in generation logic for a Bot Management feature file causing many Cloudflare services to be affected.

M

morpig•11/21/25, 10:42 AM

wondering if anyone here also have p99 >500ms for reads in same region?

Original message was deleted

H

Hello, I’m Allie!•11/21/25, 8:28 PM

What is the “external resource” in this case? Do you need an HTTP API? Do you want filesystem access?

Original message was deleted

H

Hard@Work•11/21/25, 9:50 PM

You could expose a Worker that mimics the API you want?

R

Rapid•11/28/25, 10:30 PM

default sorting by ID is extremely weird btw. I think you guys should change that to Name

PPerry Sahnow Is there a way to search through KV stores for specific metadata? Eg. to search ...

1

1984 Ford Laser•12/4/25, 9:35 PM

Searching by prefix would be the closest you can get for this

K

khazit•12/9/25, 7:45 PM

I found a similar (unanswered) question from 2024, but I'll try anyway: I need to implement a counter on Workers using KV, but it has a 1 write per second key limit (https://developers.cloudflare.com/kv/platform/limits/). Is the limit really enforced? Is there any workaround?

Kkhazit I found a similar (unanswered) question from 2024, but I'll try anyway: I need t...

H

Hard@Work•12/9/25, 7:47 PM

What exactly are you trying to do?

HHard@Work What exactly are you trying to do?

K

khazit•12/9/25, 7:50 PM

A "usage counter" within the worker, where a local meter aggregates the local counter and flushes the collected data periodically

Kkhazit A "usage counter" within the worker, where a local meter aggregates the local co...

H

Hard@Work•12/9/25, 7:50 PM

You could use a DO, which aggregates counts, and then periodically pushes to KV?

K

khazit•12/9/25, 7:52 PM

I could, but KV is simpler as I already use it for other things on the same worker

K

khazit•12/9/25, 7:53 PM

So it is enforced and there is no workaround?

Kkhazit So it is enforced and there is no workaround?

H

Hard@Work•12/9/25, 7:53 PM

I will note too that KV Read TTL is 60 seconds, so even if you were able to write multiple times a second, any readers would see updates a minute later.

TL;DR no

K

khazit•12/9/25, 7:54 PM

Ok thanks

M

mackong•12/15/25, 11:42 AM

Hi, I’m getting a 429 error when writing to Workers KV via the HTTP API. The docs say the Enterprise plan allows unlimited writes to different keys. I’m on Enterprise and writing to different keys. How can I resolve this?

Mmackong Hi, I’m getting a 429 error when writing to Workers KV via the HTTP API. The doc...

T

thomasgauvin•12/15/25, 6:55 PM

Are you writing more than 1 time per second per key? That is a hard limit

Tthomasgauvin Are you writing more than 1 time per second per key? That is a hard limit

M

mackong•12/16/25, 2:55 AM

Thanks for the reply. The RESP API limits requests to 1,200 every five minutes.

Mmackong Hi, I’m getting a 429 error when writing to Workers KV via the HTTP API. The doc...

I

Isaac McFadyen•12/16/25, 2:57 AM

I believe you'll need to reach out to your CSM to get this limit relaxed, the part in the docs about it being unlimited on Enterprise is specifically regarding the Workers binding and the REST API is a seperate limit unrelated to KV.

IIsaac McFadyen I believe you'll need to reach out to your CSM to get this limit relaxed, the pa...

M

mackong•12/16/25, 3:28 AM

Yes, I’ve switched to using a worker to write to KV, but I see there’s no bulk write support. How can I write to KV concurrently from a worker? My code currently looks like

async function putKeys(env: Env, pairs: Array<Record<string, string>>): Promise<Response> {
  for (const pair of pairs) {
    var lastE = null;

    for (let i = 0; i < retries; i++) {
      try {
        await env.KV_NS.put(pair["key"], pair["value"], {
          metadata: {
            version: "v1",
            timestamp: ((Date.now() / 1000) | 0)
          }
        });
        console.log({"message": `Successful to sync key ${pair["key"]}`});
        break;
      } catch(e) {
        lastE = e;
      }
    }

    if (lastE != null) {
      console.error({"message": `Failed to sync key ${pair["key"]}`, error: lastE});
    }
  }
  console.log({"message": `Successful synced ${pairs.length} pairs`});

  return new Response(null, {});
}

async function putKeys(env: Env, pairs: Array<Record<string, string>>): Promise<Response> {
  for (const pair of pairs) {
    var lastE = null;

    for (let i = 0; i < retries; i++) {
      try {
        await env.KV_NS.put(pair["key"], pair["value"], {
          metadata: {
            version: "v1",
            timestamp: ((Date.now() / 1000) | 0)
          }
        });
        console.log({"message": `Successful to sync key ${pair["key"]}`});
        break;
      } catch(e) {
        lastE = e;
      }
    }

    if (lastE != null) {
      console.error({"message": `Failed to sync key ${pair["key"]}`, error: lastE});
    }
  }
  console.log({"message": `Successful synced ${pairs.length} pairs`});

  return new Response(null, {});
}

async function putKeys(env: Env, pairs: Array<Record<string, string>>): Promise<Response> {
  for (const pair of pairs) {
    var lastE = null;

    for (let i = 0; i < retries; i++) {
      try {
        await env.KV_NS.put(pair["key"], pair["value"], {
          metadata: {
            version: "v1",
            timestamp: ((Date.now() / 1000) | 0)
          }
        });
        console.log({"message": `Successful to sync key ${pair["key"]}`});
        break;
      } catch(e) {
        lastE = e;
      }
    }

    if (lastE != null) {
      console.error({"message": `Failed to sync key ${pair["key"]}`, error: lastE});
    }
  }
  console.log({"message": `Successful synced ${pairs.length} pairs`});

  return new Response(null, {});
}

async function putKeys(env: Env, pairs: Array<Record<string, string>>): Promise<Response> {
  for (const pair of pairs) {
    var lastE = null;

    for (let i = 0; i < retries; i++) {
      try {
        await env.KV_NS.put(pair["key"], pair["value"], {
          metadata: {
            version: "v1",
            timestamp: ((Date.now() / 1000) | 0)
          }
        });
        console.log({"message": `Successful to sync key ${pair["key"]}`});
        break;
      } catch(e) {
        lastE = e;
      }
    }

    if (lastE != null) {
      console.error({"message": `Failed to sync key ${pair["key"]}`, error: lastE});
    }
  }
  console.log({"message": `Successful synced ${pairs.length} pairs`});

  return new Response(null, {});
}

Mmackong Yes, I’ve switched to using a worker to write to KV, but I see there’s no bulk w...

I

Isaac McFadyen•12/16/25, 3:29 AM

You probably want to use standard Promise concurrency: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Promise/all

Note that Workers can only make 6 requests at once (that includes KV I believe) so you may need to use something like https://www.npmjs.com/package/p-limit

IIsaac McFadyen You probably want to use standard Promise concurrency: https://developer.mozilla...

M

mackong•12/16/25, 3:41 AM

Thank you very much

M

mackong•12/17/25, 3:52 AM

I’m getting a 403 when listing regional hostnames in a zone, even with DNS read/write permissions. Which additional permission is required?

M

mackong•12/17/25, 3:53 AM

The code is

const client = new Cloudflare({
  apiToken: CLOUDFLARE_API_TOKEN
});

console.log("List all Regional Services regions available for use by this account.");
for await (const regionListResponse of client.addressing.regionalHostnames.regions.list({
  account_id: ACCOUNT_ID,
})) {
  console.log(regionListResponse);
}

// Got `error: 403 {"result":null,"success":false,"errors":[{"code":1002,"message":"forbidden"}],"messages":[]}`
console.log("List all Regional Hostnames within a zone.");
for await (const regionalHostnameListResponse of client.addressing.regionalHostnames.list({
  zone_id: ZONE_ID
})) {
  console.log(regionalHostnameListResponse);
}

const client = new Cloudflare({
  apiToken: CLOUDFLARE_API_TOKEN
});

console.log("List all Regional Services regions available for use by this account.");
for await (const regionListResponse of client.addressing.regionalHostnames.regions.list({
  account_id: ACCOUNT_ID,
})) {
  console.log(regionListResponse);
}

// Got `error: 403 {"result":null,"success":false,"errors":[{"code":1002,"message":"forbidden"}],"messages":[]}`
console.log("List all Regional Hostnames within a zone.");
for await (const regionalHostnameListResponse of client.addressing.regionalHostnames.list({
  zone_id: ZONE_ID
})) {
  console.log(regionalHostnameListResponse);
}

const client = new Cloudflare({
  apiToken: CLOUDFLARE_API_TOKEN
});

console.log("List all Regional Services regions available for use by this account.");
for await (const regionListResponse of client.addressing.regionalHostnames.regions.list({
  account_id: ACCOUNT_ID,
})) {
  console.log(regionListResponse);
}

// Got `error: 403 {"result":null,"success":false,"errors":[{"code":1002,"message":"forbidden"}],"messages":[]}`
console.log("List all Regional Hostnames within a zone.");
for await (const regionalHostnameListResponse of client.addressing.regionalHostnames.list({
  zone_id: ZONE_ID
})) {
  console.log(regionalHostnameListResponse);
}

const client = new Cloudflare({
  apiToken: CLOUDFLARE_API_TOKEN
});

console.log("List all Regional Services regions available for use by this account.");
for await (const regionListResponse of client.addressing.regionalHostnames.regions.list({
  account_id: ACCOUNT_ID,
})) {
  console.log(regionListResponse);
}

// Got `error: 403 {"result":null,"success":false,"errors":[{"code":1002,"message":"forbidden"}],"messages":[]}`
console.log("List all Regional Hostnames within a zone.");
for await (const regionalHostnameListResponse of client.addressing.regionalHostnames.list({
  zone_id: ZONE_ID
})) {
  console.log(regionalHostnameListResponse);
}

M

mackong•12/17/25, 11:29 AM

Is there a way to force immediate KV propagation so the edge worker reads the latest value?

Mmackong Is there a way to force immediate KV propagation so the edge worker reads the la...

H

Hello, I’m Allie!•12/17/25, 4:57 PM

No

Mmackong Is there a way to force immediate KV propagation so the edge worker reads the la...

T

thomasgauvin•12/17/25, 9:38 PM

@mackong why do you want this? please share you use case so we can understand how to prioritize

Tthomasgauvin <@1074712113991401543> why do you want this? please share you use case so we can...

M

mackong•12/18/25, 2:34 AM

We use KV for application configuration. Most settings rarely change, but for certain data—such as a status change from A to B—workers must see the update immediately.

T

thomasgauvin•12/18/25, 2:42 PM

Can you elaborate on which type of your configs need changes reflected immediately? You can DM me as well if you prefer

Tthomasgauvin Can you elaborate on which type of your configs need changes reflected immediate...

M

mackong•12/19/25, 5:19 AM

Our product is paid. We store each user’s overdue status in KV; if they’re overdue, we return an error. After a user pays, the status is updated in central storage and synced to KV, but workers may read a stale overdue value and still return an error. We need a way for workers to read the latest status so users can access the product immediately after payment.

M

Mrxbox98•12/19/25, 6:24 AM

On the KV

getWithMetadata

getWithMetadata

getWithMetadata

getWithMetadata function, what are the possible return values for the

cacheStatus

cacheStatus

cacheStatus

cacheStatus property. I don't see any info on it here: https://developers.cloudflare.com/kv/api/read-key-value-pairs/#getwithmetadata-method

Cloudflare Docs

Read key-value pairs

To get the value for a given key, call the get() method of the KV binding on any KV namespace you have bound to your Worker code:

Mmackong Our product is paid. We store each user’s overdue status in KV; if they’re overd...

P

PhoenixSpace•12/19/25, 2:01 PM

If a 60 second TTL is too long this sounds like an eventually consistent store is the wrong product over something immediately consistent like D1 or a DO

T

thomasgauvin•12/19/25, 2:43 PM

Our product is paid. We store each

PPhoenixSpace If a 60 second TTL is too long this sounds like an eventually consistent store i...

T

thomasgauvin•12/19/25, 2:44 PM

Configuration of this type is something we aim to support, and we are looking into providing more controls over eventual consistency. We will have more to share in the new year

R

Ruz•12/22/25, 2:10 PM

hey, I recently got an email that I surpassed my KV worker limit for the free tier, but when checking the usage on my dashboard they weren't showing any numbers within the metrics, is this a known bug or is there another way to see specific kv worker details?

nevermind, i needed to check for list_complete before, otherwise its not alwaysreturning the cursor

Similar Threads

Similar Threads

Similar Threads