The Load Balancer was unable to find a placement for one or more of the Service's Replicas

2
votes

The problem with deployment. Everything is okay in develop cluster, but fails to deploy successfully in on-premises cluster.

Initially I had 3-nodes cluster and got this warning very often, then I added two more nodes and amount of failed replicas has reduced, but not disappeared.

I'm confused with the messages of this warning, can anybody explain what does it mean?

The Load Balancer was unable to find a placement for one or more of the Service's Replicas: fabric:/Dev.Fabric/CommentsCacheActorService Primary Partition f2b6415c-a69d-4447-a4b0-74cad27e5692 could not be placed, possibly, due to the following constraints and properties:
TargetReplicaSetSize: 1 Placement Constraint: N/A Depended Service: N/A

Constraint Elimination Sequence: ServiceTypeDisabled/NodesBlockListed eliminated 5 possible node(s) for placement -- 0/5 node(s) remain.

Nodes Eliminated By Constraints

ServiceTypeDisabled/NodesBlockListed Nodes must not have ServiceType Disabled or be BlockListed due to Node's Pause/Deactivate Status:

FaultDomain:fd:/dc4/r0 NodeName:vm3 NodeType:NodeType0 NodeTypeName:NodeType0 UpgradeDomain:UD3 UpgradeDomain: ud:/UD3 Deactivation Intent/Status: None/None FaultDomain:fd:/dc3/r0 NodeName:vm2 NodeType:NodeType0 NodeTypeName:NodeType0 UpgradeDomain:UD2 UpgradeDomain: ud:/UD2 Deactivation Intent/Status: None/None FaultDomain:fd:/dc2/r0 NodeName:vm1 NodeType:NodeType0 NodeTypeName:NodeType0 UpgradeDomain:UD1 UpgradeDomain: ud:/UD1 Deactivation Intent/Status: None/None FaultDomain:fd:/dc1/r0 NodeName:vm0 NodeType:NodeType0 NodeTypeName:NodeType0 UpgradeDomain:UD0 UpgradeDomain: ud:/UD0 Deactivation Intent/Status: None/None FaultDomain:fd:/dc5/r0 NodeName:vm4 NodeType:NodeType0 NodeTypeName:NodeType0 UpgradeDomain:UD4 UpgradeDomain: ud:/UD4 Deactivation Intent/Status: None/None

It is very strange because this warning goes away with time. Probably it is because of rebalancing. Any ideas about what constraint is mentioned?

1
azure-service-fabric

1 Answers

1
votes

You are right about rebalancing part. Here is an excerpt from Cluster Resource Manager Management Integration article.

In each of these conditions you’ll see a health report from the Cluster Resource Manager that provides information to help you determine what is going on and why the service can’t be placed. We call this process the “Constraint Elimination Sequence”. During it, the system walks through the configured constraints affecting the service and records what they eliminate. This way when services aren’t able to be placed, you can see which nodes were eliminated and why.

It comes down to your constraint types and priorities.

How do you resolve it? Well, this article might help. It talks about making changes in Windows registry to address heap size. Since I don't have this issue at the moment, I am not inclined to apply changes. Let me know if it works out for you.