All Products
Search
Document Center

Hologres:Common issues

Last Updated:Mar 26, 2026

Use this page to identify common Hologres issues by their error messages and symptoms, then go to the relevant troubleshooting guide.

Out of memory (OOM)

Error: Total memory used by all existing queries exceeded memory limitation.

In most cases, a query has exceeded the instance memory limit. See Troubleshoot OOM issues.

Performance degradation

If queries or data writes are slower than expected, optimize SQL queries by following the guide that matches your scenario:

High CPU or long-running queries

If sustained high CPU load is affecting your online business, or metrics show a query has been running far longer than expected, a superuser can view and cancel active queries from the Active Query page in HoloWeb. See the "Use HoloWeb to view and manage active queries in a visualized way" section in Manage queries.

Connection errors

Errors:

  • terminating connection due to idle state timeout

  • terminating connection due to idle-session timeout

  • remaining connection slots are reserved for non-replication superuser connections

These errors indicate that the instance has hit its connection limit or a connection has timed out. As a superuser, manage the number of connections, close connections that do not meet your expectations, or adjust the connection timeout period. See Manage connections.

SQL errors

Errors: bigint out of range · division by zero

These errors point to a SQL logic issue in the query. See Hologres SQL FAQ.

Permission errors

Errors: permission denied · cloud authentication failed · you have no privilege

Troubleshoot based on the specific error message. See Permissions FAQ.

MaxCompute foreign table errors

Errors:

  • Failed to get MaxCompute table:Not enable schema evolution

  • Failed to get MaxCompute table:Not enable acid table

  • ERROR: not an ORC file

See MaxCompute FAQ.

Flink integration issues

Errors: ERPC CONNECTION CLOSED · Read timed out

If Flink reads from or writes to Hologres are failing or degraded, see Flink/Blink troubleshooting.

Metrics

Use instance metrics to understand what is happening inside Hologres — including which queries are consuming the most resources and whether performance has changed over time. See Metrics.