Hi, Is there an easy way to check currently running queries and all their metrics? For instance, which query is running, which user executed it, and what are the CPU/RAM usage details and etc?

SHOW RUNNING QUERIES

does not work, and even when it does, it does not show enough information.

Audit logs

only show finished queries.

PROCESSLIST

also shows only finished queries. Only

SHOW PROC '/global_current_queries';

seems to work, but I still don’t see the running scripts there.😞

Need to @ Rocky mate. @Rocky Hi, Is there an easy way to check currently running queries and all their metrics? For instance, which query is running, which user executed it, and what are the CPU/RAM usage details and etc?

SHOW RUNNING QUERIES

does not work, and even when it does, it does not show enough information.

Audit logs

only show finished queries.

PROCESSLIST

also shows only finished queries. Only

SHOW PROC '/global_current_queries';

seems to work, but I still don’t see the running scripts there.😞

Checking currently running queries with their resource metrics in StarRocks is a common requirement for troubleshooting "heavy" queries. You are correct that

Audit logs

only show completed queries, and

SHOW RUNNING QUERIES

is specifically designed for environments with Query Queues enabled. Here are the most effective ways to see what is running right now along with CPU/RAM details: 1. The Best "Real-Time" Metric View:

SHOW PROC

The most detailed way to see CPU/RAM for active queries is using

SHOW PROC

. * For the current FE:

SHOW PROC '/current_queries';

* For the entire cluster (v3.4+):

SHOW PROC '/global_current_queries';

This view provides: * MemoryUsage: Current RAM consumption. * CPUTime: Cumulative CPU time spent. * ScanBytes/ScanRows: Data volume being processed. * ConnectionId: Use this to find the SQL script (see next step). 2. Linking Metrics to SQL Scripts:

SHOW FULL PROCESSLIST

As you noted,

SHOW PROC

lacks the actual SQL text. To see the running script, you must link the

ConnectionId

from the output above to the

Id

in the processlist:

sql
-- Use FULL to see the entire SQL statement
SHOW FULL PROCESSLIST;

* How to read it: Look for rows where the

Command

is

Query

and the

State

is not

OK

(usually it says the name of the operation). The

Info

column contains the running SQL script. * Tip: If you have many connections, match the

Id

from this list to the

ConnectionId

you found in

SHOW PROC '/global_current_queries'

. 3. For Shared-Data (Warehouse) Users:

information_schema

If you are using the StarRocks Warehouse feature (Shared-Data architecture), there is a much more convenient system table that combines both:

sql
SELECT * FROM information_schema.warehouse_queries WHERE STATE = 'RUNNING';

This table includes the

QUERY

column (SQL text) alongside the

QUERY_ID

and timing information in a single row. 4. Why

SHOW RUNNING QUERIES

might not work This command only returns results if Query Queues are enabled (

enable_query_queue = true

). It is intended to show which queries are waiting in the queue (

PENDING

) versus which are actually executing (

RUNNING

). If the feature is off, the command typically returns an empty set or an error. Summary Checklist: 1. Check Metrics:

SHOW PROC '/global_current_queries';

(Look for high

MemoryUsage

or

CPUTime

). 2. Get SQL Script:

SHOW FULL PROCESSLIST;

(Match the

ConnectionId

to the

Id

). 3. Kill a runaway query: If you find a query eating too many resources, use

KILL QUERY <connection_id>;

(using the ID from the processlist). References • using_starrocks/running_queries.md • sql-reference/information_schema/warehouse_queries.md • administration/management/resource_management/query_queues.md • sql-reference/sql-statements/cluster-management/nodes_processes/SHOW_RUNNING_QUERIES.md