GCP Networking Timeout
Incident Report for Edlink
Starting around 8:30am CST this morning, Google Cloud Platform ("GCP"), Edlink's cloud computing infrastructure provider, began having an issue with Google Cloud Networking. This caused network timeouts for some GCP customers, one of which was Edlink. The network problems affected GCP's Iowa data center and as a result, Edlink periodically lost connection with our Toronto database. Edlink has one key request that must talk to our Toronto database: the part of the log-in flow where end-users are asked to enter their email address (so that we can lookup the end-user's account).

This timeout issue was causing some logins to fail randomly after trying to load for 30 seconds or more.

Despite the random errors, end-users were still able to log in if they simply continued trying, which, in and of itself, causes no harm to Edlink's ability to maintain our API uptime.

We pushed an improvement around 10:15am CST that will now only query the region in which your application has districts. This means if you are an organization that serves no Canadian districts, Edlink will not search our Toronto database for an account for your end-users and as such we'll be able to avoid the issue altogether. This improvement will be an ongoing part of our query process to avoid issues similar to this one in the future.

This issue did not affect any end-users:
1) who have ever signed in before and clicked on their remembered account
2) launched via LTI in the United States
3) launched via ClassLink or Clever
4) launched by standard launch URL from the Edlink Dashboard

To read more about the issue according to Google: https://status.cloud.google.com/incidents/FRpfyvgG3MUeSTTuX1Dx
Posted Jan 10, 2023 - 15:00 UTC