google-compute-engine"/>

I've created an instance group that is serving as the back-end for a load balancer. Steady-state for the instance group is 2 machines, with autoscaling turned on at 75%.

On 4/13/20, at 12:56 PM ET, a compute.instanceGroupManager.updateHealth event was logged, and immediately the instance group started instance re-creation for both of the currently running instances. There is no indication that the instances were ever actually unhealthy before the re-creation started.

Fast forward to today (4/20/20), at 12:55 PM ET, a compute.instanceGroupManager.updateHealth event was logged, and immediately the instance group started instance re-creation for both of the currently running instances. As before, there is no indication that the instances were ever actually unhealthy before the re-creation started.

There's no indication that it's doing a rolling replacement of the boxes that are up and running, because the same 2 instance names are being reused, and no new boxes are created to take the load while the existing boxes are being recreated. The updateHealth events do not contain any useful data about what they are actually supposed to indicate (unless the jsonPayload.resource.type value being invalid is significant). The only real commonality between the two events is being exactly one week apart, to within 2 minutes.

I've looked through the resources I've been able to search up online, and I don't see anything whatsoever that would obviously indicate that this is normal behavior.

Is this expected default behavior for an instance group?

Answer
      • 1
    • I think for this kind of behavior is to open a support case with google, then it will be easy to get the answer, however it seems this activity log entries will be discontinued, with no replacement, on June 1, 2020

Warm tip !!!

This article is reproduced from Stack Exchange / Stack Overflow, please click

Trending Tags

Related Questions