The ability to take my money for unlimited flow execution time! I faced multiple errors because of it.
Everything about Inventor has been great, except for these execution time limits…
The ability to take my money for unlimited flow execution time! I faced multiple errors because of it.
Everything about Inventor has been great, except for these execution time limits…
Unlimited time won’t be added. We don’t limit execution time to cause problems, it’s needed for resource management. The platform would have serious reliability issues if we allowed anyone to use unlimited compute.
Our goal is/has always been to extend maximum execution time over time, as we’ve been able to better manage resource allocation.
Additional time may be available in the future, but we’d rather make continual improvements and bundle it with plans rather than making it an upcharge.
Another thing that has been considered is a way for a certain number of flows (per minute, hour, or day) to run over time, for things like regular jobs. That might become available once async folders are rewritten/updated.
Yeah, I feel like maybe this feature could have gotten implemented in sooner. I’m almost willing to spend more just to have a feature like this somehow, especially since I’ve had errors because of execution time limits, so perhaps it could be added. PLEASE!
When time limit exceeds we need to bypass by executing the rest after it finishes. Won’t it be the same amount of resources or even more? Because it executes after to finish the job. At the end of the day, I think that it’s actually extending the usage, as it needs to run multiple times to finish a single job, so I guess, more resources being used for longer time periods.
Not really, because you are limited in the number of concurrent flows, and are limited in the ways flows can be spawned.
Again, for the specific use of longer running jobs, I’m interested in offering a better solution here, but we will not offer unlimited execution on all flows.
Interesting, how would be that solution?
Here’s one of the flows I have that was exceeding this limit, so I had to create a bypass before migrating to a server:
Exceeds 45sec at 85 rows.
That’s a DB perf issue, not a reason to need to bypass flow execution time limits.
If you can provide some concrete numbers (how long the lookup took, rows returned, etc.) that would be helpful.
There’s no error, but Network Request times… At that time I was sending the data to Google Sheets through Network Request at the end of each loop (db row), if I’m not mistaken, it was sending between 20-25 requests per execution limit.
Maybe ignoring time consumed by Network Requests can be a solution, as they are already limited by minute and are the blocks that consume the most time because of response waiting. You tell me.
We’ve considered switching to measuring just CPU time and limiting that to ~10ms. However, we aren’t currently set up to easily meter that so it would be a pretty involved change. And we’d likely still meter overall time to prevent abuse through extremely long running network requests, but that could be in the 3-5 minute range. Another option is potentially limiting block executions per flow, but I think people would dislike that.
Isn’t 10ms cpu time very low or am I missing something?
No, if anything that is quite generous and most flows likely finish up orders of magnitude more quickly. CPU time - Wikipedia
Regardless, after some looking, CPU time isn’t something we can easily measure because of how Inventor splits up flow execution.