Does somebody has experience with Elasticsearch on Openstack?
During our tests we notice that sometimes a node is freezed for some secondes when the Hypervisor has a problem and the node is moved to another Hypervisor.
In such cases the cluster status becomes "red" and there are always some shards unassigned. They can be assigned manually, but I would like to prevent this behaviour.
Does somebody notice a similar behaviour and was able to fix it? Maybe the connection or timeout parameters of ES can be changed to consider a freezed node in a cluster as "down".
Thanx in advance!