I am collecting Jira events and currently the moment the event happens it is written to Forge storage. When referring to the documentation, I see that this is not recommended, because there is a 1000 write operations / 20 seconds quota.
Therefore I am following the example using Async Events, where the storage.set operations are delayed. However there is another quota of ~ 1000 invocations per user per week.
Do these queue consumer calls contribute to the invocation quota?
To confirm this is the documentation you’re referring to?
There’s an initial 50,000 invocations for the first 100 seats. Following is an additional 500 invocations per seat thereafter.
Is this higher invocation number suitable for your needs?
Thanks for the reply! Yes, I’m referring to that documentation.
It seems the number is quite low (if the instance where the app is installed has more than 100 users) and I’m worried that if 1 event triggers 1 invocation the allowed weekly invocations won’t be sufficient.
It’s not definitive but we’ve received few complaints from developers reaching this limit so far.
With what little comfort that brings.
Thank you! As you can imagine, this isn’t very reassuring. Is there an API or tool that allows developers to monitor these metrics in real time? I know the developer console provides some insights, but it’s quite limited and doesn’t display runtime metrics at all.
I had a chat with the team owning the developer console.
Metrics are accurate to 2-3 minutes. Given the large timescale and number of invocations that should be sufficiently real time. Let me know if you disagree of course.
There is a metrics tab where the number of invocations in a week can be viewed. Additionally you can filter these invocations on the installation they’re coming from.
Let me know if there are problems with the above.
Thank you for clarifying, @JoshuaHwang !
It’s good that there is a way to monitor the invocations, but do you know how the runtime could be monitored? I can’t imagine an adequate way of doing it if there are no tools or APIs provided by Forge.
Is this what you’re looking for?
Otherwise I’d recommend creating logs that measure portions of your code and using those to derive numbers.