Do any v3 dependencies contact a Jovo server or API that has been affected by the v4 update?


#1

I’m getting an error from my lambda function where the code hasn’t changed in a few months. Neither has my own API that the lambda function is contacting:
“error”: {
“type”: “INVALID_RESPONSE”,
“message”: “An exception occurred while dispatching the request to the skill.”
}
Could that be because of changes to the API that affect v3?

Thanks!


#2

I haven’t been having any problems with v3 that seem related to the introduction of v4. Check your console log to see what exception is being thrown; if necessary, add a try/catch which prints the exception to the log before rethrowing it.

The main problem I’ve had with lambdas has been getting them to respond quickly enough not to time out. Adjusting the timeout on the lambda itself helps but really shouldn’t be necessary. The biggest problem in my code appears to be happening when the lambda process is being spun up (for the first time, or after having been terminated because it was idle). It’s rumored that doing a Javascript compression (mild obfuscation, turning everything into single-character symbols) may help; eliminating unused libraries may also help. And I may need to change a just-in-time data retrieval to run as a separate scheduled process.

(May have just found my problem: Update loop confusion between local time and UTC. I knew that was going to be an issue and tried to cheat. Self-inflicted wounds.)


#3

There were no API changes to v3, so the launch shouldn’t have any effects on v3 apps (except when you’re developing locally, take a look at using both CLIs in parallel).

Could you share a request and response JSON?


#4

The actual issue seemed to be close to what Keshlam had: one of the API endpoints that the Lambda function was calling had started throwing gateway timeout errors (haven’t ascertained yet what caused this). Jan seems to be right - we didn’t have to change anything in our Jovo code nor update any dependencies for it.
The issue just went away after redeploying the code to our API (without any changes). Didn’t have to increase the Lambda function timeout either.
Thank you both for your help!