Getting 404 errors when connecting to paused sandboxes

[emclaughlin215]

Describe the bug
When resuming a sandbox using .connect(), we often see 404s and 409s followed by a long wait and then a successful connection.

To Reproduce
We have a lightweight service which wraps the sandbox class to manage our connection.

This is implemented in a few places where we want to either create a new sandbox or connect to an existing one which may or may not be paused.

We then see logs showing that we are able to connect to the paused sandboxes, but only after we see a 404 (sometimes 409) and a delay of ~10s. Presumably this is due to the sandbox waking up.

Expected behavior
We connect to a paused sandbox, it takes some time (hopefully not 10s) and then we have access to the sandbox.

[linear]

E2B-2999 Getting 404/409 errors when connecting to paused sandboxes

[emclaughlin215]

Some additional logs. Looks like when connecting to a paused sandbox using .connect() and defining a timeout, the flow is something like:

1. try to add timeout to sandbox
2. error because sandbox is paused so cannot be found
3. resume sandbox

My (possibly faulty) understanding of the timeout parameter is that the sandbox would first be resumed and then have its timeout set to the provided value.

[jakubno]

Hey @emclaughlin215, sorry for these error logs; they're in a way expected. I will check if we can suppress them.

[emclaughlin215]

@jakubno Thanks! We are also seeing 409s frequently when connecting to a sandbox. Could this also be to do with the resuming logic?

[emclaughlin215]

@jakubno Did you manage to look into this anymore? Both the 404s and 409s.

[emclaughlin215]

@jakubno Any progress on this?

[jakubno]

Hey @emclaughlin215, sorry for late answer. We'll improve this, so there are no error logs, but currently there are tasks with higher priority. Both of these are expected and handled in our SDK. You can suppress the logs by adding

logging.getLogger("e2b.api").disabled = True