All Products
Search

This Product

    E-MapReduce:Check items and key metrics of Hive

    Document Center

    E-MapReduce:Check items and key metrics of Hive

    Last Updated:Mar 26, 2026

    This document covers the health check items and key monitoring metrics for HiveServer and HiveMetaStore in E-MapReduce (EMR) clusters.

    Severity levels

    Level Meaning Action required
    P0 (Critical) Hive service is unavailable Troubleshoot immediately
    P1 (High) Hive service is available but performance is degraded or workload is elevated Troubleshoot immediately

    HiveServer check items

    inspection_hive_server_availability

    Checks whether HiveServer is available by executing a set of validation statements.

    Error message Cause Action
    hive server availability permission check is failed The user lacks permissions to run HiveServer check statements (for example, permissions were accidentally revoked) Restore the required permissions
    Hive server availability is failed HiveServer is abnormal Check the HiveServer process and logs

    inspection_hive_server_ha

    Checks the high availability (HA) status of all HiveServer components.

    Result message Status Severity Action
    Hive server HA status is OK All HiveServer components are normal — No action needed
    One or more Hive server failed One or more HiveServer components are abnormal P1 Check the HiveServer process and logs
    All Hive server are failed All HiveServer components are abnormal P0 Check the HiveServer process and logs

    inspection_hive_server_port

    Checks whether port 10000 of HiveServer exists on the host. If port 10000 does not exist, the HiveServer process is abnormal. Check the HiveServer process and logs.

    inspection_hive_server_gc

    Checks the memory.heap.usage metric of the Java virtual machine (JVM) for HiveServer.

    JVM heap memory usage Severity Action
    ≥ 95% P0 Increase HiveServer memory immediately. Otherwise, HiveServer may restart and running jobs may fail.
    ≥ 90% P1 Increase HiveServer memory immediately. Otherwise, HiveServer may restart and running jobs may fail.
    < 90% Normal Adjust memory based on business requirements if needed.

    inspection_hive_server_restart

    Monitors HiveServer restart frequency within any five-minute window.

    Restart behavior Severity Action
    Repeated restarts within five minutes P0 Check the HiveServer process and logs immediately
    One restart within five minutes P1 Check the HiveServer process and logs immediately
    No restarts Normal No action needed

    HiveMetaStore check items

    inspection_hive_metastore_ha

    Checks the high availability (HA) status of all HiveMetaStore components.

    Result message Status Severity Action
    Hive metastore HA status is OK All HiveMetaStore components are normal — No action needed
    One or more metastore failed One or more HiveMetaStore components are abnormal P1 Check the HiveMetaStore process and logs immediately
    All Hive metastore are failed All HiveMetaStore components are abnormal P0 Check the HiveMetaStore process and logs immediately

    inspection_hive_metastore_port

    Checks whether port 9083 of HiveMetaStore exists on the host. If port 9083 does not exist, the HiveMetaStore process is abnormal. Check the HiveMetaStore process and logs immediately.

    inspection_hive_metastore_gc

    Checks the memory.heap.usage metric of the JVM for HiveMetaStore.

    JVM heap memory usage Severity Action
    ≥ 95% P0 Increase HiveMetaStore memory immediately
    ≥ 90% P1 Increase HiveMetaStore memory immediately
    < 90% Normal Adjust memory based on business requirements if needed.

    inspection_hive_metastore_restart

    Monitors HiveMetaStore restart frequency within any five-minute window.

    Restart behavior Severity Action
    Repeated restarts within five minutes P0 Check the HiveMetaStore process and logs immediately
    One restart within five minutes P1 Check the HiveMetaStore process and logs immediately
    No restarts Normal No action needed

    Key metrics of HiveServer2

    View these metrics on the Monitoring tab of your cluster in the EMR console.

    Category Metrics What to look for
    Session OpenSessions, ActiveSessions A spike in open or active sessions at the time of an error can indicate memory pressure. Adjust memory based on your business requirements.
    JVM JVM MemHeapMax, garbage collection (GC) metrics Review JVM metrics in the time window when errors occurred to determine whether to increase memory.
    Task PENDING tasks, ActiveRunTasksCalls, TasksCount metrics A large number of pending tasks may indicate insufficient HiveServer memory, YARN resource queue contention, or large jobs consuming most available resources. For example, the pending tasks can be the tasks whose progress is suspended.

    Key metrics of HiveMetaStore

    View these metrics on the Monitoring tab of your cluster in the EMR console.

    Category Metrics What to look for
    JVM JVM MemHeapMax, GC metrics Review JVM metrics in the time window when errors occurred to determine whether to increase memory.
    Metadata operations GetTable-related metrics, CreateTable Time A steady increase in these metrics or an exception indicates a memory bottleneck in HiveMetaStore or a performance issue in the backend database. Increase HiveMetaStore memory based on the current memory size, or upgrade the backend database specifications based on the time required to run a query on the database.