ElasticSearch can take some time to get ready for queries and actual activation code do not wait for it's readiness.
If a query is sent to ES right after activation there's a chance it's not ready and one gets the following error: "Failed to execute phase [query], all shards failed".
This error pops up quite often when using embedded ES in unit tests.
To mitigate this we should, on activation:
- ensure the Qi4j index is refreshed
- wait for ES yellow status (the primary shard is allocated but replicas may not be)