Hi Prefect Friends, I have been using prefect for ...
# prefect-cloud
j
Hi Prefect Friends, I have been using prefect for the past 2 years; in two companies, with very different workloads... In our current implementation we are running a Flow every 20s (which is much more frequent than I have in the past). However, we seem to get a gap every hour that lasts 10 minutes (5:30, 4:23) as you can see in this flow runs: Is this expected behaviour? Are we at some threshold?
Any help would be appreciated 🙏
r
interesting that the tags are different after the gap
👀 1
in that its missing auto-scheduled
can you filter the flow runs screen to that flow and future scheduled runs - is there a gap in them?
j
Ya, thanks for noticing. let me see if there is something there...
That seems to be the only one. Maybe that was just someone kicking it off manually? I am not sure. Here are some other gaps: 3 minutes gap
r
any gaps in the future scheduled ones?
j
I am going to turn it on shortly; we turned it off as it was skipping for 10m. I am just reviewing with another engineer to see how many gaps over the past 24 hours. It actually doesn't look as bad as we thought. This happened in the first couple hours... give me a couple minutes... there is a lot of entries 🙂
😁 1
r
any concurrency limits in play?
j
No, not for this. In this case we are using Prefect for scheduling, and calling an HTTP endpoint (our internal endpoint) which does 'the work'. So the load isn't an issue. The only big difference for us is the frequency. As these are running 3 times a minute. Everything else we are doing; is every >15 minutes.
Workers, aren't scaled with a lot of headroom for both the Prefect Worker and our Web application (API).
(thats as far back as I am going to go)
The other engineer has a headache, so she is taking the next few hours off. I won't enable the schedule until we are caught up together. Sorry for the delay 🙏 When we get back together, I will enable the schedule and look for future gaps.
👍 1