|
const timeoutMs = this.config.requestTimeoutMs ?? 120000; |
The timeout for the WS transport is set at 2 minutes.
This means if you run a process via execStream in the Sandbox, and it takes over 2 minutes (streaming from an agent, for example), the stream dies with an unhandled error timeout after 120000ms.
This also appears to be an issue when I start a process as a background process, and then connect to its log stream, presumably same code path.
Suggestion: allow this to be configurable in wrangler file and document that the default is only 2mins.
alternatively, if I'm missing something let me know!