Before you begin reading this guide, we recommend you run Elasticsearch Error Check-Up which can resolve issues that cause many errors.
This guide will help you check for common problems that cause the log ” Unexpected error during the primary phase for action ; request ” to appear. To understand the issues related to this log, read the explanation below about the following Elasticsearch concepts: replication and request.
Advanced users might want to skip right to the common problems section in each concept or try running the Check-Up to analyze Elasticsearch configuration and help resolve this error.
Overview
Replication refers to storing a redundant copy of the data. Starting from version 7.x, Elasticsearch creates one primary shard with a replication factor set to 1. Replicas never get assigned to the same node on which primary shards are assigned, which means you should have at least two nodes in the cluster to assign the replicas. If a primary shard goes down, the replica automatically acts as a primary shard.
What it is used for
Replicas are used to provide high availability and failover. A higher number of replicas is also helpful for faster searches.
Examples
Update replica count
PUT /api-logs/_settings?pretty
{
"index" : {
"number_of_replicas" : 2
}
}Common problems
- By default, new replicas are not assigned to nodes with more than 85% disk usage. Instead, Elasticsearch throws a warning.
- Creating too many replicas may cause a problem if there are not enough resources available in the cluster.
Log Context
Log “unexpected error during the primary phase for action [{}]; request [{}]” classname is TransportReplicationAction.java.
We extracted the following from Elasticsearch source code for those seeking an in-depth context :
assert false : "finishAsFailed called but operation is already finished";
}
}
void finishWithUnexpectedFailure(Throwable failure) {
logger.warn("unexpected error during the primary phase for action [{}]; request [{}]"; failure; actionName; request);
if (finished.compareAndSet(false; true)) {
setPhase(task; "failed");
listener.onFailure(failure);
} else {
assert false : "finishWithUnexpectedFailure called but operation is already finished";
Try AutoOps to detect and fix issues in your cluster:
