does lumenize handle setWebSocketAutoResponse under the hood? couldnt find it mentioned in the docs

SSilvan does lumenize handle setWebSocketAutoResponse under the hood? couldnt find it me...

L

Larry•11/17/25, 11:30 AM

I believe Lumenize RPC does, although I can't remember. Lumenize has gotten pretty big and I have three things that use WebSockets and can't remember which set the auto response pair.

SSilvan does lumenize handle setWebSocketAutoResponse under the hood? couldnt find it me...

L

Larry•11/17/25, 11:33 AM

Correction. It does not.

// Note: DOs can set up their own auto-response via this.ctx.setWebSocketAutoResponse()
// in their constructor if needed. We don't set one here to avoid overwriting custom
// auto-responses. RPC clients implement their own heartbeat via setKeepAlive().

// Note: DOs can set up their own auto-response via this.ctx.setWebSocketAutoResponse()
// in their constructor if needed. We don't set one here to avoid overwriting custom
// auto-responses. RPC clients implement their own heartbeat via setKeepAlive().

// Note: DOs can set up their own auto-response via this.ctx.setWebSocketAutoResponse()
// in their constructor if needed. We don't set one here to avoid overwriting custom
// auto-responses. RPC clients implement their own heartbeat via setKeepAlive().

// Note: DOs can set up their own auto-response via this.ctx.setWebSocketAutoResponse()
// in their constructor if needed. We don't set one here to avoid overwriting custom
// auto-responses. RPC clients implement their own heartbeat via setKeepAlive().

LLarry Correction. It does not. ```typescript // Note: DOs can set up their own auto-re...

S

SilvanOP•11/17/25, 11:33 AM

alright perfect thanks!

Llambrospetrou I don't know about Cap'n Web, but in normal Workers RPC you have e-order pipelin...

L

Larry•11/17/25, 11:37 AM

Interesting. Lumenize RPC awaits each thing in the batch before executing the next one so it's guaranteed in order. Again though, I'm removing batching. I think chaining and nesting are more in line with how RPC should be used. Like what you wrote before, if you need a batch, create a method and call it.

H

HatKid•11/17/25, 12:03 PM

just made something with durable objects! it's good apart from the fact that you have to build it directly from your folder, and can't via the web editor. Any solution? i generally build on-the-go so I generally don't have access to wrangler.

HHatKid just made something with durable objects! it's good apart from the fact that you...

K

Kevin•11/17/25, 12:54 PM

you can use something like github codespaces or just use github code editor and actions to deploy to cloudflare

Llambrospetrou Side note: you call it "batches" in the Luminize docs, but we call it "RPC promi...

L

Larry•11/17/25, 1:41 PM

Also, a batch might imply some sort of transaction-ish guarantees, which does not exist here.

For Lumenize RPC it is transaction-ish. Batches are a single round trip and they all occur sychronously (all are awaited).

LLarry > Also, a batch might imply some sort of transaction-ish guarantees, which does ...

L

lambrospetrou•11/17/25, 2:50 PM

But, storage operations are not in a transaction right? If my chain modifies storage, two writes for example, and the 2nd one fails, then the 1st one will still be persisted.

BTW, I am not arguing what Lumenize provides is not useful, just arguing that it can get very confusing with these tricks. I know they are interesting technically, but I usually find that it's a very nice way to introduce bugs too.

Llambrospetrou But, storage operations are not in a transaction right? If my chain modifies sto...

L

Larry•11/17/25, 7:43 PM

I'm not certain, which is why I'm leary of counting on it. If you await a synchronous storage call, is it still held by output gates? Is the await ignored? Also confirm my understanding that if you do NOT await them, then two writes are held by output gates. The either both succeed or both fail. Is that correct?

KKevin you can use something like github codespaces or just use github code editor and ...

H

HatKid•11/17/25, 8:56 PM

Noted. Thanks!

LLarry I'm not certain, which is why I'm leary of counting on it. If you await a synchr...

L

lambrospetrou•11/17/25, 9:31 PM

Also confirm my understanding that if you do NOT await them, then two writes are held by output gates. The either both succeed or both fail. Is that correct?

I did a detailed explanation of this few months ago, but the way I think about it now is simply:

split all your statements by
```
await
```
await
```
await
```
await
IO.
anything between
```
await
```
await
```
await
```
await
IO runs as an implicit transaction.

So, in your question, yes, assuming the code is running INSIDE the DO, two storage writes without any await in between them will run in an implicit transaction and either they both will persist or none.

Llambrospetrou > Also confirm my understanding that if you do NOT await them, then two writes a...

L

Larry•11/17/25, 9:44 PM

That doesn't quite answer the question. Does

await

await

await

awaiting a non-async storage operation count as

await

await

await

awaiting IO? Or would it only count if you were awaiting a true async method call? Or even

await

await

await

awaiting the older async kv storage operations. I remember a conversation with Kenton years ago where he indicated that they were special and didn't open output gates.

LLarry That doesn't quite answer the question. Does `await`ing a non-async storage oper...

C

cloudkite•11/17/25, 10:10 PM

Why await something that’s synchronous?

C

cloudkite•11/17/25, 10:11 PM

If the DO output gate mechanism is based on the js run loop then seems like await anything would trigger it. Just speculating

Ccloudkite If the DO output gate mechanism is based on the js run loop then seems like awai...

L

Larry•11/17/25, 11:27 PM

You'd think but I clearly remember Kenton saying they altered that the JS engine level for

await

await

await

awaiting the old async kv storage operations. It stands to reason they would also do that for the sync storage, but not a given.

S

see translation•11/17/25, 11:52 PM

I had a project that was using the DO KV API. But I want to use the the new SQL API.

Running my tests with vitestI got an error like:

"SQL is not enabled for this Durable Object class. To enable it, change

new_classes

new_classes

new_classes

new_classes to

new_sqlite_classes

new_sqlite_classes

new_sqlite_classes

new_sqlite_classes within the 'migrations' field in your wrangler
.jsonc or wrangler.toml file. If using workerd directly,set

enableSql = true

enableSql = true

enableSql = true

enableSql = true in your workerd config for the class. Note that this change cannot be made after the class is already deployed to production"

So to work around this I renamed my classes and changed all the corresponding names. As well as the "name" in the wrangler.toml but when I tried to run the tests again I get the same error. What do I have to change to make a "new" deployment?

Ssee translation I had a project that was using the DO KV API. But I want to use the the new SQL...

L

Larry•11/18/25, 12:01 AM

Once a DO is in production, you can't change it. However, if it you are just working locally in vitest-pool-workers you can change your wrangler.toml to your heart's content.

S

see translation•11/18/25, 12:02 AM

vitest is giving me these errors when I run my worker locally to test. I can't figure out how to change my wrangler.toml to make it work.

S

see translation•11/18/25, 12:03 AM

Is there a way I can like just deploy to a second new deployment?

L

Larry•11/18/25, 12:03 AM

Might you have more than one wrangler.jsonc or wrangler.toml file?

L

Larry•11/18/25, 12:04 AM

You could post your file here for us to see if we can spot something.

S

see translation•11/18/25, 12:05 AM

name = "sql-sync-service"
main = "build/index.js"
compatibility_date = "2025-11-17"

[build]
command = "cargo install -q worker-build && worker-build --release"

[durable_objects]
bindings = [
  { name = "SYNC_SESSION", class_name = "SyncSession" },
]

[[migrations]]
tag = "v1"
new_sqlite_classes = ["SyncSession"]

[observability.logs]
enabled = true

[observability.traces]
enabled = true
head_sampling_rate = 1
persist = true

# DEFAULT ENV -- Used in Local Development & Debugging Remote
[[d1_databases]]
binding = "USER_PEER_MAPPING"
migrations_dir = "./database/user-peer-mapping/migrations/"
database_name = "user-peer-mapping"
database_id = "REDACTED"


[vars]
INTERNAL_API_SECRET_KEY = "LOCAL_API_KEY"
SPS_URL = "https://REDACTED"

[[kv_namespaces]]
binding = "DOCUMENT_VERSIONING_KV"
id = "REDACTED"

[[kv_namespaces]]
binding = "SNAPSHOT_STORE_KV"
id = "REDACTED"

name = "sql-sync-service"
main = "build/index.js"
compatibility_date = "2025-11-17"

[build]
command = "cargo install -q worker-build && worker-build --release"

[durable_objects]
bindings = [
  { name = "SYNC_SESSION", class_name = "SyncSession" },
]

[[migrations]]
tag = "v1"
new_sqlite_classes = ["SyncSession"]

[observability.logs]
enabled = true

[observability.traces]
enabled = true
head_sampling_rate = 1
persist = true

# DEFAULT ENV -- Used in Local Development & Debugging Remote
[[d1_databases]]
binding = "USER_PEER_MAPPING"
migrations_dir = "./database/user-peer-mapping/migrations/"
database_name = "user-peer-mapping"
database_id = "REDACTED"


[vars]
INTERNAL_API_SECRET_KEY = "LOCAL_API_KEY"
SPS_URL = "https://REDACTED"

[[kv_namespaces]]
binding = "DOCUMENT_VERSIONING_KV"
id = "REDACTED"

[[kv_namespaces]]
binding = "SNAPSHOT_STORE_KV"
id = "REDACTED"

name = "sql-sync-service"
main = "build/index.js"
compatibility_date = "2025-11-17"

[build]
command = "cargo install -q worker-build && worker-build --release"

[durable_objects]
bindings = [
  { name = "SYNC_SESSION", class_name = "SyncSession" },
]

[[migrations]]
tag = "v1"
new_sqlite_classes = ["SyncSession"]

[observability.logs]
enabled = true

[observability.traces]
enabled = true
head_sampling_rate = 1
persist = true

# DEFAULT ENV -- Used in Local Development & Debugging Remote
[[d1_databases]]
binding = "USER_PEER_MAPPING"
migrations_dir = "./database/user-peer-mapping/migrations/"
database_name = "user-peer-mapping"
database_id = "REDACTED"


[vars]
INTERNAL_API_SECRET_KEY = "LOCAL_API_KEY"
SPS_URL = "https://REDACTED"

[[kv_namespaces]]
binding = "DOCUMENT_VERSIONING_KV"
id = "REDACTED"

[[kv_namespaces]]
binding = "SNAPSHOT_STORE_KV"
id = "REDACTED"

name = "sql-sync-service"
main = "build/index.js"
compatibility_date = "2025-11-17"

[build]
command = "cargo install -q worker-build && worker-build --release"

[durable_objects]
bindings = [
  { name = "SYNC_SESSION", class_name = "SyncSession" },
]

[[migrations]]
tag = "v1"
new_sqlite_classes = ["SyncSession"]

[observability.logs]
enabled = true

[observability.traces]
enabled = true
head_sampling_rate = 1
persist = true

# DEFAULT ENV -- Used in Local Development & Debugging Remote
[[d1_databases]]
binding = "USER_PEER_MAPPING"
migrations_dir = "./database/user-peer-mapping/migrations/"
database_name = "user-peer-mapping"
database_id = "REDACTED"


[vars]
INTERNAL_API_SECRET_KEY = "LOCAL_API_KEY"
SPS_URL = "https://REDACTED"

[[kv_namespaces]]
binding = "DOCUMENT_VERSIONING_KV"
id = "REDACTED"

[[kv_namespaces]]
binding = "SNAPSHOT_STORE_KV"
id = "REDACTED"

S

see translation•11/18/25, 12:06 AM

In my tests I setup miniflare like:

export async function setupMiniflare() {
  const mf = new Miniflare({
    d1Databases: {
      USER_PEER_MAPPING: userPeerMappingDatabaseID,
    },
    scriptPath: "./build/worker/shim.mjs",
    modules: true,
    modulesRules: [
      { type: "CompiledWasm", include: ["**/*.wasm"], fallthrough: true },
    ],
    durableObjects: {
      SYNC_SESSION: "SyncSession",
    },
    r2Buckets: {
      DOCUMENT_SNAPSHOT_BUCKET: "document-snapshot-bucket",
    },
    kvNamespaces: {
      DOCUMENT_VERSIONING_KV: "document-versioning-kv",
      SNAPSHOT_STORE_KV: "snapshot-store-kv-test",
    },
    cachePersist: false,
    workflowsPersist: false,
    durableObjectsPersist: false,
    bindings: {
      DOCUMENT_PERMISSIONS_SECRET: "local",
      INTERNAL_API_SECRET_KEY: "INTERNAL_API_SECRET",
      INTERNAL_API_SECRET,
      SPS_API_SECRET_KEY: "local",
      SPS_URL:"http://localhost:8090",
      local:true,
    },
    compatibilityDate: '2025-11-17'
  });

export async function setupMiniflare() {
  const mf = new Miniflare({
    d1Databases: {
      USER_PEER_MAPPING: userPeerMappingDatabaseID,
    },
    scriptPath: "./build/worker/shim.mjs",
    modules: true,
    modulesRules: [
      { type: "CompiledWasm", include: ["**/*.wasm"], fallthrough: true },
    ],
    durableObjects: {
      SYNC_SESSION: "SyncSession",
    },
    r2Buckets: {
      DOCUMENT_SNAPSHOT_BUCKET: "document-snapshot-bucket",
    },
    kvNamespaces: {
      DOCUMENT_VERSIONING_KV: "document-versioning-kv",
      SNAPSHOT_STORE_KV: "snapshot-store-kv-test",
    },
    cachePersist: false,
    workflowsPersist: false,
    durableObjectsPersist: false,
    bindings: {
      DOCUMENT_PERMISSIONS_SECRET: "local",
      INTERNAL_API_SECRET_KEY: "INTERNAL_API_SECRET",
      INTERNAL_API_SECRET,
      SPS_API_SECRET_KEY: "local",
      SPS_URL:"http://localhost:8090",
      local:true,
    },
    compatibilityDate: '2025-11-17'
  });

export async function setupMiniflare() {
  const mf = new Miniflare({
    d1Databases: {
      USER_PEER_MAPPING: userPeerMappingDatabaseID,
    },
    scriptPath: "./build/worker/shim.mjs",
    modules: true,
    modulesRules: [
      { type: "CompiledWasm", include: ["**/*.wasm"], fallthrough: true },
    ],
    durableObjects: {
      SYNC_SESSION: "SyncSession",
    },
    r2Buckets: {
      DOCUMENT_SNAPSHOT_BUCKET: "document-snapshot-bucket",
    },
    kvNamespaces: {
      DOCUMENT_VERSIONING_KV: "document-versioning-kv",
      SNAPSHOT_STORE_KV: "snapshot-store-kv-test",
    },
    cachePersist: false,
    workflowsPersist: false,
    durableObjectsPersist: false,
    bindings: {
      DOCUMENT_PERMISSIONS_SECRET: "local",
      INTERNAL_API_SECRET_KEY: "INTERNAL_API_SECRET",
      INTERNAL_API_SECRET,
      SPS_API_SECRET_KEY: "local",
      SPS_URL:"http://localhost:8090",
      local:true,
    },
    compatibilityDate: '2025-11-17'
  });

export async function setupMiniflare() {
  const mf = new Miniflare({
    d1Databases: {
      USER_PEER_MAPPING: userPeerMappingDatabaseID,
    },
    scriptPath: "./build/worker/shim.mjs",
    modules: true,
    modulesRules: [
      { type: "CompiledWasm", include: ["**/*.wasm"], fallthrough: true },
    ],
    durableObjects: {
      SYNC_SESSION: "SyncSession",
    },
    r2Buckets: {
      DOCUMENT_SNAPSHOT_BUCKET: "document-snapshot-bucket",
    },
    kvNamespaces: {
      DOCUMENT_VERSIONING_KV: "document-versioning-kv",
      SNAPSHOT_STORE_KV: "snapshot-store-kv-test",
    },
    cachePersist: false,
    workflowsPersist: false,
    durableObjectsPersist: false,
    bindings: {
      DOCUMENT_PERMISSIONS_SECRET: "local",
      INTERNAL_API_SECRET_KEY: "INTERNAL_API_SECRET",
      INTERNAL_API_SECRET,
      SPS_API_SECRET_KEY: "local",
      SPS_URL:"http://localhost:8090",
      local:true,
    },
    compatibilityDate: '2025-11-17'
  });

also FYI my worker is written in rust

S

see translation•11/18/25, 12:06 AM

but my tests are js/ts

L

Larry•11/18/25, 1:16 AM

When you said vitest, I thought meant that you were using vitest-pool-workers not miniflare. I'm guessing the problem is that miniflare is more like production and keeps files around that can't be converted. If it's anything like

wrangler dev

wrangler dev

wrangler dev

wrangler dev, those files are in .wrangler folder. If you clear that, you should be good to go.

S

sathoro•11/18/25, 1:25 AM

Has anyone had the 50 GB DO KV storage account limit increased?

I've submitted three tickets over several months now and it's getting beyond frustrating

The latest response to my new ticket is from my account manager trying to get me to hop on a call with a product manager

Please Cloudflare just increase the limit I don't want to hop on a call for a simple storage limit increase its ridiculous

Not to mention I don't live in those time zones so I'm not available and the email from the account manager isn't even referring to the correct product she started talking about R2... It's just getting ridiculous

S

sathoro•11/18/25, 1:34 AM

Also does anybody know what happens when you hit the limit? I assume write operations start failing?

We are at 48GB now

Ssathoro Has anyone had the 50 GB DO KV storage account limit increased? I've submitted ...

X

xtudos•11/18/25, 1:44 AM

It seens that will be best if you move to sqlite rather then keeping to wait this limit to be increased

Xxtudos It seens that will be best if you move to sqlite rather then keeping to wait thi...

S

sathoro•11/18/25, 1:52 AM

For our use case KV is much better and it would be prohibitive to migrate them at this point

S

sathoro•11/18/25, 2:01 AM

KV itself is unlimited so I really don't understand this limit at all... Seems like it's just an arbitrary limit for them to try to upsell you or something. We are already on the Startup Enterprise Plan and with the terrible support we get (like an AI closing the last ticket without a human ever seeing it) it's not really selling us on the "Enterprise" support being worth it

Ssathoro For our use case KV is much better and it would be prohibitive to migrate them a...

X

xtudos•11/18/25, 2:01 AM

You could create a table to store
Key value in another DO

And inside the current DO you make the logic to write in the new DO

X

xtudos•11/18/25, 2:02 AM

So you keep the current one workiing
And dont have to worry about have to migrate all of them rn

X

xtudos•11/18/25, 2:02 AM

Just like R2 recommends doing when migrating from another cloud bucket

Ssathoro KV itself is unlimited so I really don't understand this limit at all... Seems l...

X

xtudos•11/18/25, 2:03 AM

Kv is cached across all cloudflare datacenters

But not DO
It lives in a single location

S

sathoro•11/18/25, 2:03 AM

Huh? Why would writing in a new DO help? It's an account level limit

X

xtudos•11/18/25, 2:04 AM

Because its only limited on KV

X

xtudos•11/18/25, 2:04 AM

In the new DO you would write in a sql

S

sathoro•11/18/25, 2:05 AM

Yeah I understand that's a solution but our most storage intensive DO is quite complex with the way we use the KV storage and would be very difficult to actually do that. I appreciate the help but I really am not needing you to suggest workarounds

X

xtudos•11/18/25, 2:06 AM

You could just make an interface , so for the current code, it wouldn't actually change

X

xtudos•11/18/25, 2:08 AM

Because you are risking going the entire application down
At least now you have some time left to do the changes

S

sathoro•11/18/25, 2:11 AM

Got it, thanks

LLarry When you said vitest, I thought meant that you were using vitest-pool-workers no...

S

see translation•11/18/25, 2:31 AM

I tried that. still no luck. I run the tests with

npx vitest run

npx vitest run

npx vitest run

npx vitest run

Ssee translation I tried that. still no luck. I run the tests with `npx vitest run`

L

Larry•11/18/25, 2:58 AM

Maybe miniflare stores them somewhere else? When you say you tried that, what did you try?

Ssathoro For our use case KV is much better and it would be prohibitive to migrate them a...

1

1984 Ford Laser•11/18/25, 3:17 AM

Are you aware that you can use the KV API in a SQL backed DO class? I understand this would still mean you would need to migrate the existing instances but better than some other solution imo

Xxtudos Kv is cached across all cloudflare datacenters But not DO It lives in a single ...

M

mars•11/18/25, 5:25 AM

interesting

Xxtudos Kv is cached across all cloudflare datacenters But not DO It lives in a single ...

I

Isaac McFadyen•11/18/25, 5:39 AM

KV is cached across all colos. If a key is not retrieved from a specific colo then it will never be stored there - the actual storage only occurs in 2 core locations (a place in the US and in the EU).

IIsaac McFadyen KV is *cached* across all colos. If a key is not retrieved from a specific colo ...

L

lambrospetrou•11/18/25, 8:06 AM

Several things mentioned here.

First of all, sathoro above is referring to the KV backed DOs, not the Workers KV product.
And the KV backed DOs are NOT replicated across all DCs, and not cached across all DCs either. The DO KV-backed storage is still only in the region of the DO.

Are you aware that you can use the KV API in a SQL backed DO class?

As another user said, you could use the KV API in the new SQLite-backed DOs, although the per DO storage there is not unlimited, it's still 10GB hard limit but there is no account limit, so you would need to a) manually copy the data over, b) shard the data more.

Llambrospetrou Several things mentioned here. First of all, sathoro above is referring to the ...

S

sathoro•11/18/25, 8:14 AM

Our data is already sharded across thousands of DOs so the issue isn't the per instance storage it's the hard account-level limit which just feels weird and arbitrary. They're asking me things like why do you need the limit increased? I don't know how to respond to that other than well we are about to hit the limit and we need to store more data...

S

sathoro•11/18/25, 8:14 AM

This is with Enterprise support by the way

Ssathoro Our data is already sharded across thousands of DOs so the issue isn't the per i...

L

lambrospetrou•11/18/25, 8:18 AM

Yeah, I will ask the team internally too and see if I get some clarification about that limit.
It used to be just to avoid abusers, so the question is more of a sanity check that you are legit user, but considering your contract it should be simple ask.

does lumenize handle setWebSocketAutoResponse under the hood? couldnt find it mentioned in the docs

Similar Threads

Similar Threads

Similar Threads