Hi.
I try to fetch killmails with id and hash from zkillboard today.
The status of my http requests are something like this:
200 [success]
200 [success]
200 [success]
200 [success Erh … I think that means I didn’t make any mistake.]
420 [Suddenly, I get a 420!]
420 [“This software has exceeded the error limit for ESI …”]
420 [I don’t get this.]
420 [Ok. You win. I give up.]
Why would my program hit the error limit when all requests are successful?
Setting a user-agent string for my program doesn’t solve the problem.
Is this normal?
Does this have something to do with the blackout of ESI bookmarks?
Did CCP set a rate limit when they upgrade the API?
In short you need to handle the error limit, by respecting the headers: x-esi-error-limit-remain = how many fails you have left in the time frame set by reset
and x-esi-error-limit-reset = how many seconds until the remain is reset to max value
The error limit is global for all endpoints.
Once remain hit zero, you will get the 420. So, once remain hit zero (or get close to the limit - if you have multiple requests in flight), you need to stop all requests for reset seconds.