As you can see the basic resources info for lxd0 node,
you’re running out of CPU capacity because the total cpu capacity for the lxd0 = 16 * 4 -> 64. And for now there are 32 containers in the ams, and each of them has instance-type
a2.3 assigned, which occupied 32 * 2 = 64 vCPU cores. That’s the reason why there are no more cpu resources to be allocated for a new session.
As long as you deleted all
stopped containers, a new session could be get launched again. Please read the doc for more details.
In general, after a session is terminated, those
stopped containers should be deleted by AMS automatically, that’s a bug in 1.18.x release. The fix will be landed in 1.19.0 release.