im just wondering if its redundant
im just wondering if its redundant
HIt shouldn't be? Though actually, it might error if you don't provide it?
1
1so it isnt redundant
HYeah
1thanks
BIf you generate and query by name, you have to supply the jurisdiction. If you use durable object IDs, the jurisdiction is embedded in the ID so you don't have to limit the namespace. This has been my experience
Chey folks, supposing I wanted to bind a sandbox to an internal service (separate DO), is that possible? I was digging around the docs and didn't find anything too promising, but wanted to double check here.
basically I want to be able to do something like:
worker → create sandbox, execute commands in a loop → sandbox-backend (handles things like auth)
feel free to also redirect me if there's a better channel
basically I want to be able to do something like:
worker → create sandbox, execute commands in a loop → sandbox-backend (handles things like auth)
feel free to also redirect me if there's a better channel
H?crossposting
SPlease do not post your question in multiple channels/post it multiple times per the rules at #
welcome-and-rules. It creates confusion for people trying to help you and doesn't get your issue or question solved any faster.
welcome-and-rules. It creates confusion for people trying to help you and doesn't get your issue or question solved any faster.
LThe Worker to DO pattern is the norm, but I'm uncertain what you mean by "sandbox". Is it untrusted code?
1Probably the CF Sandbox SDK https://developers.cloudflare.com/sandbox/ which is basically containers with a wrapper
LAhh, makes sense. I'm unfamiliar with that product, although I think it would be cool to get my AI Coding Agent to run in one of them. Just not sure how I'd see the output.
1Stream it back via WebSockets
LWould need something like that. I'm sure someone has already built that or will soon.
MRequest/Question:
I was wondering if there’s a way to perform a cross join between data stored in different Durable Object tables — similar to how PostgreSQL uses foreign data wrappers (FDWs). It would be game changer.
I was wondering if there’s a way to perform a cross join between data stored in different Durable Object tables — similar to how PostgreSQL uses foreign data wrappers (FDWs). It would be game changer.
LThat would be a neat trick, but I've never heard of anything like that, and I probably would have if it existed.
AIt's built in, you can open a ws to anything insider your sandbox
https://developers.cloudflare.com/sandbox/guides/websocket-connections/
https://developers.cloudflare.com/sandbox/guides/websocket-connections/
LIt's good to know that the WebSocket part is already built in. Do you know anyone who is using it as a remote AI coding platform? I'd love to hear more about the setup.
ANo I just recently found out about this myself. Looks like a killer offering tho... docker container, with DO-like ergonomics
ADoes anyone happen to know, if you call a DurableObject via RPC, does that implicitly also wake up its associated Worker?
To ask differently. Do I get billed more if I do:
CallerA -> WorkerB.rpc() -> DO-B.rpc()
vs
CallerA -> DO-B.rpc()
To ask differently. Do I get billed more if I do:
CallerA -> WorkerB.rpc() -> DO-B.rpc()
vs
CallerA -> DO-B.rpc()
LWhat is CallerA? Is it client-side (browser or app)? Workers RPC is between Workers, DOs, and RpcTargets. If you want all the way to the client, you need something like Cap'n Web or Lumenize RPC. If you are just using Workers RPC, the called DO is accessed directly without the involvement of the Worker in the module where that DO is defined.
Khow can I view storage.kv/_cf_KV items for remote durable objects?
SHihi! If I spin up a Durable Object to store session data via the KV API and give the data an expiry via TTL, will the Durable Object still exist after the data's expired? If so, is there any residual wrapper that might incur storage costs?
LWorkers KV has expiration dates but the KV inside a DO does not. You'd have to implement that yourself, or use Workers KV.
LYou can add a method on the remote DO that returns it and then call that method remotely using Workers RPC. So,
if you are using id's instead of instance names its:
The only tricky part most of the time is knowing the id or name of the remote DO.
this.env.REMOTE_DO.getByName('remote-instance-name').get_cf_KV(). If you are going to call the remote DO more times, I recommend that you two-step it and put the the remote DO stub in a variable. So:if you are using id's instead of instance names its:
The only tricky part most of the time is knowing the id or name of the remote DO.
Sthanks for the info. in terms of how to implement in DO, is this only doable via alarms or are there other alternatives?
1that wouldve saved me a lot of work lol....
1at least it looks consistent... i added jurisdiction even when using the DO IDs 
HYou can nuke the
.wrangler directory, or find the specific DO namespace folder and delete it
THI everyone,
is it possible to mix a durableObject to have Sqllite and also key value?
i have some structured data that i want to save and avoid recalculating, the issue that if i use tabular structure its take for ever to save to sqlite, and if i use a json its failing cause this json is too big.
what is the best practice here?
is it possible to mix a durableObject to have Sqllite and also key value?
i have some structured data that i want to save and avoid recalculating, the issue that if i use tabular structure its take for ever to save to sqlite, and if i use a json its failing cause this json is too big.
what is the best practice here?
LAs of 2025-09-12 compatiblity date you can do
const x = this.ctx.storage.kv.get(”x”); // no await
JHi, I was also wondering the same thing...
I am not a developer but an infra guy ... could you please let me know how to dump SQLlite manually?
I am not a developer but an infra guy ... could you please let me know how to dump SQLlite manually?
LExport for backup is a real hole in Cloudflare's offering at this time. I keep saying, "You have read replicas working. Please, just disconnect one and stick the .sqlite file in R2." The response I keep getting is that idea is just one option that's part of a larger discussion. What I haven't heard is that I'm wrong and it's not as easy as I think, but I do wonder if that's true.
LWhat I haven't heard is that I'm wrong and it's not as easy as I think, but I do wonder if that's true.I said this multiple times
Right now, you cannot just go and "disconnect one and stick the .sqlite file in R2". It can be done, and that's our plan in some form, but it's not a 2 hour thing our next intern can do.
And to add more colour, it's not that it takes 8 months to do this, it's just that other priorities postponed this work.
LExporting, backups, migration to some other database system, are all on the roadmap. You can argue they should be higher priority than what is delivered, which is fair, but we can say that for all the other 100 features users are asking for.
LWhen I've been in product manager roles in the past, whenever some annoying person (like I am right now) tells me something should be easy, my reply is, "Not as easy as you think, but if you tell me which of these 10 other things that are near the top of the backlog you would like this to take priority over, we will put it in that spot. Think about it overnight and we'll adjust when you come back in the morning." They usually look at the list and tell me that they think the priorities are pretty close to right as-is. I'm sure that's the case now also. 
BNot just for backup, real data is also useful for debugging. It's not always simple to manually reproduce a complex database locally.
Jwith a Durable Object - is it possible to get the name you used
idFromName from within the DO?Jin my log I'm seeing:
LIt is not unless you pass it in somehow. Many people have an init() method that they call at the right moment(s) that sticks it in storage. I put the name in a header that my fetch handler plucks out of the header and populates. That only works for me because I never create new instances via Workers RPC. My instances are always created with an HTTP call (either a real fetch/Request or the WebSocket upgrade request).
JI'm basically using all RPC calls I think
Jso why does it have the
{ name : undefined }?Jlike not that I set it as undefined, what's even the point of it being there
LThe type is for
DurableObjectId, which is used both on the caller side of the RPC and inside the DO. In the caller side the .name will be populated correctly. The fact it's not populated inside the DO is a long-term known bug.Jso is there a case when it should show?