wrongly disabled used nodes

Bug #1710535 reported by suzhengwei on 2017-08-14
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
watcher
Medium
Unassigned

Bug Description

When testing server consolidation, I found that it occasionally disabled still used nodes. The reason is following.
If new instances is created during the time between audit and actionpaln executing, it has high probability to schedule the new instances into the unused node. But watcher has still disabled the node which still has the new running instances. This is what the strategies not expected .

suzhengwei (sue.sam) on 2017-08-14
description: updated
Changed in watcher:
milestone: none → queens-1
Alexander Chadin (joker946) wrote :

Watcher team have discussed it and we have found the following solution:
1. Advice to use auto-trigger option for consolidation strategy.
2. As of starting Action Plan we check for new instances.
3. Watcher consume notifications from Nova during Action Plan execution.
4. If second and third steps are true, Watcher send a notification about changing of cluster state.

Changed in watcher:
status: New → Triaged
importance: Undecided → Medium
milestone: queens-1 → rocky-2
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers