We’re getting intermittent SIP error 408 timeouts on inbound calls - request timed out waiting for a response. It started happening immediately after pushing a small change to the architect flow associated with the 1877-555-1212 number. The flow now pipes the caller ID through a data action to set a custom field - nothing fancy.
The SIP stack is on v2024.2.0. Console shows the INVITE getting a 408, but the PBX logs aren’t offering anything useful beyond that. It’s probably something changed in the SIP mediation layer after the architect rollout.
Is there a known issue with data actions in architect impacting SIP timeouts? A quick fix to bypass it would be nice.
The 408 error - request timeout - usually happens when the data action is taking too long to complete. The documentation states that data actions “execute synchronously within the context of the call flow” - this means the Architect flow waits for the data action to finish before continuing. If the data action takes longer than the SIP TTL, you’ll get this timeout.
The suggestion above is correct - increasing the timeout helps. But setting it to 3600 seconds feels high. We found in INC-4471 that a value of 60 seconds for dataActionTimeout is often enough for most data actions. A small thing - the default is 10, which is often too low.
The flow pipes the caller ID - that’s fine. But it’s the data action itself that’s likely slow. Change the dataActionTimeout setting in the data action configuration to 60. If it still fails, then we may need to look at the data action endpoint response time. The documentation recommends monitoring data action performance - a slow endpoint will cause this.
Side note - if you’re using a third-party data action, check their service status page.