Comment 21 for bug 1730717

Revision history for this message
Iain Lane (laney) wrote :

Just a heads up - this apparently became much harder for me to reproduce at will. We're still seeing it in actual workloads but I'm having trouble recreating manually.

My current strategy is to start stress-ng on a number of machines and then constantly reboot them, with the idea that this will stress the cloud and make the bug more likely, which relies on the assumption that this bug is somehow to do with busyness on the underlying machines.