I’m using a worker in the smallest tier which apparently doesn’t have enough memory (512MB) for my purposes. The process gets killed regularly due to running out of memory.
The weird thing is that the memory metrics for the VM don’t even reach the stated capacity of 512MB:
The metrics tab might be a little misleading as it is an average over three minutes for your service. Spikes in memory usage that could cause your service to be killed for using too much memory might happen too quickly for the metrics backend to report on the dashboard.
