Helix API timeout when trying to get my followers


#1

Quite a while ago I have written a script which gets my followers and exports it to a file. The script worked fine until October 11th. Since then it does not work reliable anymore, most request are simply timing out without ever returning a result. I don’t see any reason or problem on my side, I tried this from different IP addresses and via different internet connections / from different servers.

My request URL is https://api.twitch.tv/helix/users/follows?to_id=MYUSERID&first=100 (and in subsequent request with &after=CURSOR added). For a while the first two or three requests (so the initial and then one or two followups with a cursor added) went through and then it stopped working, but lately not even the first request gets a result. I also tried it with lower values for first to no avail.

So what is going on here? How can this be broken for over 1 1/2 months now?

PS: It also does not work when using v5-API.


#2

MYUSERID is not a valid User ID. So I cannot debug your specific API Call

I am unable to replicate your issue calling another UserID

How are you making this call? Did you try using Raw cURL instead of your Library in case the library is at fault?


#3

Here is the proper URL.
https://api.twitch.tv/helix/users/follows?to_id=403455113&first=100

It is not a problem of the library I am using. When trying it with wget I also experience problems.

However I think I might have identified a factor in this issue. I am located in Germany and my requests to “api.twitch.tv” end up somewhere in (probably) Frankfurt on the machine with IP 151.101.14.214. It might be a problem specific to the server there.

When I run my script from a VM in the Netherlands where api.twitch.tv resolves to 151.101.38.214 and probably ends up somewhere in Amsterdam it works just fine.


#4

Germany quite often has REALLY BAD issues connecting to Twitch in general.

So that’ll be the issue.

You can try fudging with your servers DNS and see if you can pick a better DNS provider, but the Germany issue is usually something screwed up in the routing. Which is a Twitch/Amazon can’t fix


#5

Now please, what kind of statement is this? We are not internet 3rd world here. If there are consistent problems with connections from Germany as claimed by you, it is 100% something Twitch/Amazon or the service providers used by you (e.g. Fastly is hosting the API host) can and has to fix. We have extensive and well built networks here and there are usually no problems with any service or target whatsoever. I’m sorry, but this really sounds like a cop out reply.

However your claim of general “really bad issues” does not hold true in my opinion anyway. There are tons of Twitch streamers located in Germany and we all stream without any issues 99,999% of the time. When we experience problems they are almost always due to our ISPs. The only cases where it was not like that I remember where when Twitch had bigger/global outages.

Furthermore, I have these issues both from my server in the datacenter of one of the biggest Hosters in Germany as from my home internet connection. Both have totally different connection paths to api.twitch.tv. So it really makes no sense to claim that the problem is anywhere else than in the realms of fastly or on that server.

Forcing a different IP through a override in /etc/hosts for example does not help, as the routing for the netblock still ends up at the Fastly peering in Frankfurt.

I am convinced that there is a issue with that specific peering of Fastly in Frankfurt or that specific server and that Fastly should have a look into that instead of making statements like “Germany has generally problems”. Again, we have world class internet here so claiming that this is a issue of Germany and not of Fastly or the API setup quite frankly does not make any sense at all.


#6

All I know if that quite often Germany users are known to report issues with accessing Twitch. This pops up on the dev Discord, reddit and other places periodically.

It happens quite often (granted I’ve not heard a report recently) and usualy the issue is internal to Deutsche Telekom. Not much Amazon/Twitch can do about that other than complain to them. (again this is based on reports from users on Reddit whom when comparing providers report they are using DT)

“You”? Please note I am not Twitch staff, just another third party developer.

Then the issue is with Fastly, and you probably want to take it up with them as it’s probably affecting other services, not just routing to Twitch.


#7

Deutsche Telekom is not even involved here in the slightest. And I am sorry, your reply sounded like you were a employee of Twitch and you might understand why considering the points I have brought up your reply might rub one the wrong way. I apologize for my harsh words, but I can assure you there are no general Germany-wide issues with Twitch (anymore), to my knowledge not even with Deutsche Telekom as ISP.

So are there any Twitch employees reading these forums? I don’t think I should contact Fastly myself, it is not even clear that the Issue is with Fastly or with the server hosted by them (which then would probably be the responsibility of Twitch admins), Or how to report such a issue to Twitch engineers?


#8

I have now tweeted the issue to @TwitchDevs in Twitter, it appears they are reading and handling such tweets, so wish me luck :slight_smile: