Scheduled Worker - Unexpected 429s
Hi, we're having an issue with cron scheduled worker invocations (
scheduled()) where some of our outbound fetch calls to external APIs are intermittenly getting rate limited (429s) or forbidden (403) responses .
We only make one request to these endpoints every 15 minutes, so my only guess is that this is due to other scheduled workers sharing infrastructer / egress IPs? Can someone confirm and provide guidance?2 Replies
???
Happy to provide any more information... I was handling the 429s errors in the code, so the worker and event "succeeded".
Wondering if that prevents cycling out to different infra? I changed it now to let the cron events fail... so far 3 failures in a row so no cycling of my worker.
Ok, this is now a larger question...
Given a relatively small pool of egress IP addresses used by CF, is the assumption that any sufficiently known third-party API that implements IP-based rate limiting is basically unusable from Worker functions?
At least and especially scheduled workers, which probably run on a much smaller subset of machines / IPs?
Ok. An update - When I deploy with a cron trigger defined in
wrangler.toml , the handler worker hits 429s every time - tried both * /15 * * * * and * /17 * * * * and it errors out on every invocation for hours... but if I then manually edit the cron trigger in the dashboard, it goes through without a problem.