Do you also have the problem that workflows are not triggered at irregular intervals (in active mode)?
-The trigger/workflow works in test mode
-The trigger/workflow works in normal mode (tested)
I have a workflow that is triggered when an invoice mail (Outlook) is moved to an “Invoice” folder. This also works when I try it out (in test mode and in normal mode), but I have now noticed several times that I have mails in the folder that do not trigger the workflow, but since it normally works, all settings are basically correct. It’s not as if I have an email every 10 seconds, sometimes these emails arrive with a time delay of several hours…
In the screenshots you can see that emails arrive regularly, but the workflow is hardly ever triggered.
I don’t know if this might be the case with what you are observing, but in one workflow I did against Microsoft graph API, the frequency of the workflow execution was exceeding the maximum allowed API call rate for the app/account and only the first few calls after the timer was reset (on the server end) would succeed. The only way I found to confirm whether that was the cause was to run the requests from n8n through a proxy.
I personally don’t think it’s a rate limit issue in my case. As you can see in the screenshot, the emails are coming in with large time gaps — sometimes several hours apart. Even on the more active days, the messages are spaced out enough that they shouldn’t exceed any typical API rate limits. From what I understand, rate limits usually apply when multiple requests are made within seconds or a short period, which doesn’t match my situation.
So I think the issue might be elsewhere — possibly with how the event triggers are handled on the Microsoft side or with n8n’s connection stability.
You didn’t mention in the original post whether you are self-hosted or using n8n cloud, and you didn’t mention which version of n8n. If you are running self-hosted, it may also be useful to know, assuming it’s Docker, what the Docker host environment is (mac, windows, linux, hardware, network environment, hosting company, etc.).
I don’t have any more theories based on what you’ve said so far. Just suggesting that filling in more context info might help someone reading this narrow their guesses about why it may be happening, and constrain their suggestions about how to troubleshoot.
