Downtime and Error Limiting - How are you handling it?

[Nexuscrawler]

Just a general question, because I’m interested if there’s a better way than what I’m currently doing.

I’m sending many parallel requests to ESI at a time hitting endpoints for many users. When there’s a general problem with ESI and I’m getting error codes, I’m pretty much hitting the error limit of 100 immediatly, because there are usually still over a hundred request on the fly.

When I’m getting a 420 response and am error limited, I stop all sheduling of new requests and only hit the https://esi.evetech.net/latest/status/ endpoint every 60 seconds, until it stops responding with error codes.

It works well so far, but I’m wondering how you guys are handling error limiting and downtime in particular. Would there ever be a chance that the status-endpoint would be down and the others up? As far as I know, pings to esi.evetech.net do no work. Is there a better way to check for downtime than the 503 status codes?

Thanks!

[Snitch_Ashor]

Theres two things I do. Exclude the regular daily downtime and and all cronjobs that will do a lot of calls will query the status endpoint once without retries and stop if that fails.

[Nexuscrawler]

Are you excluding the full 15 minutes of regular downtime or do you check if the servers come back online earlier?

all cronjobs that will do a lot of calls will query the status endpoint once without retries and stop if that fails.

So basically check the status endpoint before sheduling the updates? I could do that, but I don’t really have a single point where I start all the updates. I have a few timers that update different sorts of endpoints individually. Certainly would make sense though. I might just add this check to each of the timers and check for 200 and 304 codes.