I have this application that auto-migrates itself on startup if the new mapping is incompatible with the old mapping. The problem is, sometimes during a full cluster restart, it starts the reindex process while the index is in a yellow state. Is that ok? Does it make the replicate node recovery take longer or anything?
I've notice it can take quite some time for the yellow state to go away on larger indexes as well.
While in yellow state, the cluster can not guarantee the high availability requirements set by the configuration of the existing indices. Meaning that if node goes down or there is any outage, it will go red and any indexing/querying will be blocked (May also lose a shard completely, depending on severity).
But a yellow state does not means that cluster is underload. So if your reindex does not imposes too much load in the cluster and it is unlikely that it will cause any trouble, then I would say it's all good.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.