What happened?
At 2:35 PM UTC on August 3rd, we identified an issue with one of our servers failing to process Recordings. This issue lasted less than 12 minutes, and it impacted 4% of Recordings. This could've caused a small number of Recordings to be discarded.

Why did this issue occur?
The issue happened due to a temporary authentication issue with our cloud provider.

What will we do to prevent this from happening in the future?
In the future, we will hold an internal postmortem.