How to get a predictive service back up once it goes out of service?

User 481 | 9/1/2015, 6:20:11 PM

This happens way more often than expected. ps.get_status() gives a cannot connect because predictive service is OutofService error.

What are my options when this happens? How do I get the predictive service back up?

I have tried rebooting the ec2 instance, but that has not helped.

Comments

User 1174 | 9/1/2015, 8:13:31 PM

Hi MSH,

You can use ps.replace_node(['instance-id']) to replace your OutOfService instance with a fresh instance.

Thanks,


User 1174 | 9/1/2015, 8:16:51 PM

Could you elaborate on what are you trying to do with the Predictive Service? The instances should not go out of service. Can you go through some logs from the server.log or graphlab_service.log so see if there are strange errors?