Hi all- I’m looking to implement a Jira webhook handler and was wondering if anyone has some rough idea of how many requests per second to expect when a Jira cloud client is performing batch operations? If I understand, there’s no ability to have cloud Jira back off or to get logs, so I just want to design accordingly.
Best way to design is it to look at @tpettersen’s talk at AtlasCamp: https://www.atlassian.com/atlascamp/watch-sessions/advanced-atlassian-development/integration-strategies-from-atlassians-strategic-integrations-team
(Basically gather as much as you can and play it back at your own pace). That said - Webhooks are http and because of that not guaranteed - you should always do a scheduled “check to make sure if I missed anything”.
Hi @tbldave, we are using AWS Lambda to receive Jira webhooks and apply the same pattern as presented in the talk Daniel mentioned above, i.e. we receive and directly queue them. I can definitely recommend AWS Lambda for this as you get a really good & fast scalability. I have given a talk about that on Atlas Camp as well: https://www.atlassian.com/atlascamp/watch-sessions/advanced-atlassian-development/think-big-start-small-scale-fast-develop-an-app-using-aws-lambda (the webhook part is discussed at minutes 22:00 - 25:30) Let me know if you have more questions about it
@anon28347317 @danielwester Great, it makes good sense to periodically poll for any missed events and I’d definitely prefer to go the lambda and queue approach. I think your talk and these replies build a good case for going this route. Thanks!