The most probable reason is that you are running out of memory. Use htop or similar command to check on the memory. Taking a larger instance usually resolves this issue.
This means that our Kubernetes cluster is unable to allocate a pod for your job. Segmind is internally notified whenever an instance takes greater than 5 minutes (300 seconds) to boot up. Please contact us [contact(at)segmind.com] if this happens repeatedly with your account.