Probably the CF Sandbox SDK https://developers.cloudflare.com/sandbox/ which is basically containers
Probably 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
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
idFromNamein 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 }like not that I set it as undefined, what's even the point of it being there
LThe type is for
DurableObjectId.nameso is there a case when it should show?
LInside the DO, it will never show right now, until the bug is fixed internally.
Jbasically for example I have 2 DOs, each with some data in KV.
I can call
I can call
firstDO.index()secondDO.index()index()As Larry said, at the moment the workaround is to store the name inside the DO storage during the DO initialization phase if you have one, otherwise in any of the followup calls, or you pass it all the time as argument. What you do and how depends on your use-case. Cumbersome, but unfortunately these are the options to have it in there.
Jso anyone trying to fix this well known long-term bug?
Jbasically my point is if you have to pass it in to the DO - I might as well just include the data outside of the DO since that's where it's coming from anyways.
LYou can track this bug in https://github.com/cloudflare/workerd/issues/2240
JSince it is never populated inside the DO (and I suppose this won't be fixed short-term), I think it would be better and cause less confusion to have different type definitions for each place, where the one in the DO instance wouldn't have the
.name
NTrying to move a KV DO to SQLite and I know I'll have to delete it and create a new one. But how do you do that? what does the migration look like? Do I have to completely remove the binding from wrangler, build then readd it?
Ndo i simply do this?
LJust got issues with SQLite today in my DO where I stored user tables. "Error: Exceeded allowed rows written in Durable Objects free tier". Which is really strange since I'm only storing a few rows (8 users!) and 100 rows in a table called "transcripts".
JSo they just call
setIdentifierthis.identifier