Ok, how many machines are you running in parallel, and what size EC2 node are you running them on?

We are using 30 slaves with medium compute types (7gb, 4 cores)

Just to clarify: each machine has 4 cores? You’re not running 30 machines on one EC2 instance?

We use a batch build to create 30 instances (which I was calling slaves) to run the tests in parallel

Ok, to rule out lack of resources, I suggest turning off video recording and running in electron As a minimum we use 8 CPUs per machine when video recording is enabled

Yes good suggestions and we have already disabled video recording and are using electron. We are having 'luck' today and have had ~10 runs which have had no problems. Nothing has really changed on our end so I am not sure if it is solved or what. I am just waiting for a failure or another day of clean runs at this point 🤷‍♂️

Let me know if you start having an unlucky day. I spent a fair while unpicking similar weird behaviours due to resource constraints. Cypress is a hungry beast