CODEWITHSUNDEEP
google-compute-engine
Raphaël
Raphaël

Reputation: 267

compute.instances.host Error an instance during

In the operations history of my compute engine project, my machines all have an operation listed as "Automatically migrate an instance (compute.instances.automaticRestart)". They are all on the same zone, using the same debian template.

I suppose that their was some maintenance on the plateform, which is fine for me if the OS doesn't reboot.

Unfortunately two machines suffered a reboot. The operations history listed the operation as "compute.instances.hostError an instance" (compute.instances.hostError). In addition Syslog doesn't suggest a clean shutdown.

Is there anything I should/can do to prevent such problem?

edit : We are europe-west1-b and all servers have the setting : On host maintenance to Migrate VM instance

Upvotes: 6

Views: 5507

Answers (2)

kevan
kevan

Reputation: 841

Doesn't look like this was ever answered.

compute.instances.hostError means there was a hardware or software failure on the physical machine that was hosting your VM.

The FAQ has a description -- https://cloud.google.com/compute/docs/faq#hosterror

Upvotes: 8

Faizan
Faizan

Reputation: 1967

As per this article all the zones in GCE except "europe-west1-a" have Transparent maintenance where the instance will be live migrated without rebooting it. If your instance is in a zone with Transparent Maintenance you can set the option On host maintenance to Migrate VM instance using your developer console. Once this option is set your instance will be live migrated without rebooting the instance.

Upvotes: 0

Related Questions