# Ops: Aggregator Failures

## Master Aggregator Dies

If the master aggregator dies and is recoverable, all that is necessary is to restart the SingleStore process.

If it is not recoverable, then a child aggregator must be promoted to master with [AGGREGATOR SET AS MASTER](https://docs.singlestore.com/db/v7.3/reference/sql-reference/cluster-management-commands/aggregator-set-as-master.md).

## One Child Aggregator Dies

If the aggregator is recoverable, all that is necessary is to restart the SingleStore process. Otherwise, simply deploy a new SingleStore instance as an aggregator.

## Many Aggregators Die

If more than one aggregator dies, they can be reintroduced or replaced one by one.

***

Modified at: June 22, 2022

Source: [/db/v7.3/user-and-cluster-administration/dealing-with-cluster-failures/ops-aggregator-failures/](https://docs.singlestore.com/db/v7.3/user-and-cluster-administration/dealing-with-cluster-failures/ops-aggregator-failures/)

(An index of the documentation is available at /llms.txt)