As part of SEAI-100 we're using log parsing rules to detect assertions that fire on our Atlas clusters. One alert that has appeared is "Attempting to discard dirty page" on an Atlas 7.1.0 cluster.

I'll link to the Slack message in a developer only comment.

We're just now setting up the process to analyse these types of errors. Part of this ticket will be determining what information we can gather from the Atlas cluster