Investigate frequent MasterServer authentication outages

[GeckoEidechse]

For some reason in recent time master server started requiring manual restarts every 2-3 days due to starting to through auth errors on clients. Issue is currently not known and needs investigating.

[ASpoonPlaysGames]

The error clients get is the Invalid/Expired Masterserver Token

[ASpoonPlaysGames]

More details on the errors that we get [error] Northstar origin authentication failed [error] {"success":false,"error":{"enum":"STRYDER_RESPONSE","msg":"Couldn't parse stryder response"}} [error] Failed reading masterserver response: got fastify error response [error] {"success":false,"error":{"enum":"INVALID_MASTERSERVER_TOKEN","msg":"Invalid or expired masterserver token"}}

[ASpoonPlaysGames]

The error seems to be happening in one of two places: https://github.com/R2Northstar/NorthstarMasterServer/blob/main/client/clientauth.js#L64 or https://github.com/R2Northstar/NorthstarMasterServer/blob/main/client/clientauth.js#L74

So we are either failing to parse the JSON that we get back from stryder (unlikely imo) or we are getting an error back from stryder. Potentially a ratelimit of some kind?

[uniboi]

Just don't confirm ownership with stryder :trollface:

[wolf109909]

Could be. i stand on stryder rate limit too. It's most likely to happen when I just push a update to northstarCN, and around 100Players are trying to get auth from masterserver during a short period of time.

[GeckoEidechse]

This would go inside with it happening less often currently where our playerbase is on the smaller end.

(sidenote, it's shrinking roughly at the same rate as vanilla Steam playerbase post recent summer sale so not really a reason to worry)

[ASpoonPlaysGames]

Since my PR has been merged, the error I am getting has changed from

[error] Northstar origin authentication failed [error] {"success":false,"error":{"enum":"STRYDER_RESPONSE","msg":"Couldn't parse stryder response"}}

to

[error] Failed reading origin auth info response: malformed response object {"statusCode":500,"error":"Internal Server Error","message":"Cannot read properties of undefined (reading 'toString')"}

The only uses of toString that were added mean that the error is one of the two options:

1. no response from stryder, making authResponse undefined, which then makes the returning of the authResponse throw an exception
2. an exception being thrown in the asyncHttp request, also causing authResponse to be undefined, with the same result

I'm going to make a PR to add some more checks, so that I can further narrow down this issue

[Erlite]

If it's any help, I also got a 500 by passing in a + through the Origin token, while trying to see if I could inject other parameters to trick Stryder into validating my token & UID while passing a different UID to the master server.

[pg9182]

No issues for the last month after switching to Atlas.