Even so, you actually don't need to know how much each DO is consuming, you only have to use our info as reference, because you can always evict them when they hit the memory limit either way. But this way, at least we would have some form of control of what is going on. Of course the memory info would provide us much more control.

So let's use the example you gave, where you specify that each uses 10mb, and there are 10 in an isolate. What if one suddenly jumps, and hits 30mb. Do we evict specifically that one? All of them? Some randomly, until we no longer are out of memory? Without knowing accurate results, that's impossible to decide, and you'd have to evict all of them, or some randomly, even when it'd make more sense to only evict the one that's passed its limits

The issue with this is that it just makes it all more unpredictable than it is at the moment

No, because that only happens when they misbehave, right now, it might happen either way. One could use internal info to try to manage that misbehaviour, but right now, there is nothing we can do.

Of course if you want, you can even create isolates of custom sizes, not sure the overhead, but that would solve the problem.

What do you mean by nesting? What are you trying to accomplish?

I'm getting an error occasionally,

Maximum call stack size exceeded

, from my live DurableObject but not when testing via wrangler. I think it might be because I'm sending too much traffic (doing a data migration). Is that the behavior when there is too much traffic, or is that an error in user code? It happens intermittently, and it seems to be at random (i.e. another call to the same URL may work). Update: and now, after a redeploy of the worker, it seems to not be having that issue anymore. 🤔

Is there any guidance around what to do for known high request spikes in traffic? I'm migrating data from my old provider into my new DurableObjects and getting connection resets and timeouts. There isn't a lot of traffic going to each individual DurableObject, just the workers. All the traffic is coming from one location. Is there any articles about the right/best way to do this?

Are the errors coming from the DO instances, or from the Workers?

I'm not sure. I get ECONNRESET and ETIMEDOUT errors and occasionally this 1104 error. Retrying the calls works. But I want to be a good citizen and ensure if there is something I should be doing on my end that I do it. I have opened a support ticket for the errors, but I wanted to ask here if there is any guidance that I haven't read that I should be doing.

Can you send me that support ticket ID?

2754616

using websockets in DO, is there a way to paginate the connected sessions? or does the DO sessions array just hold ALL of the connections. 😦

👀

You should not be seeing script not found

I wonder if canary has something new hasnt that happened before 😛

There is something in the works on the workerd repo (as part of websocket hibernation) that will let you do that, but for the moment you will need to add all of them to an array and manage that yourself.

What's colos are you seeing this in?

What do you mean, what git repo? It is in my live worker/DO and the code for it is in a private git repo.

Oops sorry, *colos

Good question. I think DEN is the closest to me (the caller) but the DOs are close to the user around the globe. The ticket has Ray IDs for those errors if that helps.

Eng aren't able to see tickets unfortunately

That seems... counter-productive. 🙂 Can you lookup ray IDs? This is the list: 7b2b69bc8818291b 7b2b7bd728411f36 7b2b643aac41e19d 7b2b676ddf052909 7b2b7bd768a71f36 7b2b7e4e7fc31f42

I can

I'm getting the

Maximum call stack size exceeded

errors again too. I suspect if I redeploy my worker (without any changes) they will go away like before. I'm not sure if it is related to the issues, but here are some Ray Ids for that: 7b2d47536e492903-DEN, 7b2d508e9c422903-DEN, 7b2d5163e9922903-DEN

Not really seeing anything my side, no Worker release hitting the colos you're having issues, not seeing any increase in script not found errors globally. DO team may have a better idea of what is going on, I was mainly concerned of a possible Workers issue but it doesn't look like it

Thank you for checking on it. ❤️ I'll wait to see what the DO team finds.

i have router in my worker. CRUD methods are using DO. In Delete method, the state of DO is always empty. In all other methods(GET,POST and PUT), state has the latest data. can anyone help me with solution please ?

I don't quite follow. Can you elaborate more?

from the DO side of things, I didn't see anything abnormal for those rayids. from the stateless worker side of things, I do see script not found. Let me know if you don't hear anything back on that support ticket