Details
-
Sub-task
-
Status: Resolved
-
Major
-
Resolution: Won't Fix
-
None
-
None
-
None
Description
It is possible that the cluster is operating sub optimally with more nodes than required. It is not possible for Solr to automatically identify a suboptimal utilization of nodes. We expect the user to provide us with policy rules to identify it
examples:
{"cores" : ">2", "node":"#ANY" , "strict" : false } {"freedisk" : "<10", "node":"#ANY" , "strict": false}
Solr would go through all these rules to identify if it breaks any such rules. If yes, it would go through the following steps
- Identify all violations to such rules
- Sort all nodes and identify the least loaded nodes
- If there is an empty node
- store that node name
- Simulate removal of that node and go to step #1
- Check if they are violating these rules
- if yes, simulate moving them to other nodes with more load
- if it is possible, to move without breaking rules,
- store that suggestion
- Go to step #1
- If not possible, get the list of move operations (step #6.1) and collect the list of nodes to be removed ( step # 3.1)