In this post, I am sharing a script to check the list of running queries or statements which are waiting in the Resource Queues of Greenplum.

The Workload management of Greenplum means – Set priorities of database user’s queries using Resource Queues. If you create RQ with the limit of 10 concurrent queries and user is trying to execute more than 10 queries at a time then few RQ keeps few queries in the waiting mode.

Using below query, you can get the list of running queries which are in the waiting mode.

SELECT
	rolname
	,rsqname
	,pid
	,granted
	,current_query
	,datname
FROM pg_roles, gp_toolkit.gp_resqueue_status
,pg_locks, pg_stat_activity
WHERE pg_roles.rolresqueue=pg_locks.objid
	AND pg_locks.objid=gp_toolkit.gp_resqueue_status.queueid
	AND pg_stat_activity.procpid=pg_locks.pid
	AND pg_stat_activity.usename=pg_roles.rolnam

In this post, I am sharing one of the important articles on how to reset the priority of running or active statement of Greenplum.

The Work Load management of Greenplum means – define the different Resource Queues with options like priority, memory, CPU.
RQ associates with the user and all statement of this user will execute basis on specified priority setting in that RQ.

Let me give you one scenario, where I used this.

Recently, I found one long running query in our Greenplum server.
I checked the priority of that running query which was LOW, and that only one query was running at that time.

So now, I wanted to complete that query as soon as possible, so the solution was to change the priority from LOW to HIGH. – For only that running query.

I used below script and changed the priority and query got completed in next six minutes. We can use gp_adjust_priority() to change the priority of running statements.

Be careful – Below query will alter the priority of queries which are running more than 20 minutes.
So if you want to analyze the queries, you should use * and comment — gp_adjust_priority(psa.sess_id,rpb.rqpcommand,’HIGH’)

SELECT gp_adjust_priority(psa.sess_id,rpb.rqpcommand,'HIGH')
FROM gp_toolkit.gp_resq_priority_backend rpb
JOIN pg_stat_activity psa
ON rpb.rqpsession = psa.sess_id
WHERE EXTRACT(epoch from current_timestamp-psa.query_start)/60 > 20

Create an RQ with Statement Limit:
It will allow an only specified number of active statement for particular RQ.

CREATE RESOURCE QUEUE MyRQ WITH (ACTIVE_STATEMENTS=5);

Create an RQ with Memory Limit:
The Memory restriction all the queries which submitted through the RQ.

CREATE RESOURCE QUEUE MyRQ WITH (ACTIVE_STATEMENTS=5, MEMORY_LIMIT='3000MB');

Create an RQ with Query Execution Planner cost Limit:
You can set a limit of Query Execution Planner cost where some active statements are not required and N number of sessions can run until it reaches to MAX_COST.

CREATE RESOURCE QUEUE MyRQ WITH (MAX_COST=200000.0);

Set Priorities:
You can set any priority level from this list: LOW, MEDIUM, HIGH, MAX

ALTER RESOURCE QUEUE MyRQ WITH (PRIORITY=MAX);

Assign Roles to RQ:
Create:

CREATE ROLE MyRole WITH LOGIN RESOURCE QUEUE MyRQ;

Alter:

ALTER ROLE MyRole RESOURCE QUEUE MyRQ;

Remove Role from RQ:

ALTER ROLE MyRole RESOURCE QUEUE MyRQ;

General parameters:

max_resource_queues: Set the maximum number of resource queues
resource_select_only: Set ON = allow SELECT, SELECT INTO, CREATE TABLE AS SELECT, and DECLARE CURSOR, Set OFF = ALL SELECT + INSERT, UPDATE, and DELETE
max_resource_portals_per_transaction: Set the maximum number of simultaneously open cursors which will allow per transaction
stats_queue_level: Enables statistics collection on usage of resource queues
gp_resqueue_priority: This is enable by default
gp_resqueue_priority_sweeper_interval: Set the interval for recalculation of CPU Usage
gp_resqueue_priority_cpucores_per_segment: Very important, you can set number of CPU cores per segment

Parameters for Memory Utilization:

statement_mem: Runtime allocation of memory to a particular query
max_statement_mem: Only super user can set this to prevent the over-allocation of statement_mem
gp_resqueue_memory_policy: It enables Greenplum memory management feature. NONE = Memory management same as Greenplum Engine, AUTO = Controls by statement_mem
gp_vmem_protect_limit: Set the limit so that queries cannot exceed the usage of physical memory of segment host