That's if the bot is coded well. I've cut off bots from sites and services before, only to have them start to DOS because their response to an error was to just try again, immediately and forever
> On March 14th, Apollo made nearly 1 billion requests against our API in a single day, triggered in part by our system outage. After the outage, Apollo started making 53% fewer calls per day. If the app can operate with half the daily request volume, can it operate with fewer?
That's a backend server making a request that doesn't count against its rate limit as soon as it can again.
The push notification backend is described as making a request every 6 seconds for each user. If it isn't backing off correctly and delaying a requeueing of the next job, that can drastically increase the request rate.
> On March 14th, Apollo made nearly 1 billion requests against our API in a single day, triggered in part by our system outage. After the outage, Apollo started making 53% fewer calls per day. If the app can operate with half the daily request volume, can it operate with fewer?
That's a backend server making a request that doesn't count against its rate limit as soon as it can again.
The push notification backend is described as making a request every 6 seconds for each user. If it isn't backing off correctly and delaying a requeueing of the next job, that can drastically increase the request rate.