Problem Statement

Many users have reported frustration with workspaces not automatically shutting down as expected with their autostop configuration. The TTL of these workspaces is extended by our activity bump feature, which postpones the shutdown when a user is "active" in a workspace.

However, the definition of activity can be a bit convoluted, and consequently, the activity bump behavior is unpredictable. Administrators want to diagnose exactly why a workspace is being kept alive when seemingly no connections are active. Then, the administrator can hunt down the connection and close it or roll out a permanent solution.

A solution should help administrators understand:

• When the last time a workspace's TTL was bumped
• The type of connection that incremented it
• The source of that connection (some client ID possibly?)

A proposal

A user suggested logging the events or the last event that bumped activity on the workspace page. Something simple like a schedule tooltip that shows:

Activity detected from <source> <time> ago. 

Activity detected from webterminal 15 minutes ago. 

Activity detected from SSH 46 minutes ago.