How-to-deploy-postgres-exporter-on-SaaS-to-monitor-postgres-custom-query_704971984

The community repo of postgres-exporter is https://github.com/prometheus-community/helm-charts/tree/main/charts/prometheus-postgres-exporter

The image verison of postgres-expoter we used is v0.12.0. we have scaned the image security.

Preparation steps:

Create a dedicated postgres read user to access database.

CREATE USER grafana_reader WITH PASSWORD '*****';

GRANT CONNECT ON DATABASE xservices_ems TO grafana_reader;
GRANT CONNECT ON DATABASE xservices_rms TO grafana_reader;
GRANT CONNECT ON DATABASE idm TO grafana_reader;

GRANT CONNECT ON DATABASE bo_ats TO grafana_reader;

\c xservices_ems
GRANT USAGE ON SCHEMA maas_admin TO grafana_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA maas_admin TO grafana_reader;

\c xservices_rms
GRANT USAGE ON SCHEMA maas_admin TO grafana_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA maas_admin TO grafana_reader;

\c idm
GRANT USAGE ON SCHEMA idm TO grafana_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA idm TO grafana_reader;

\c postgres
GRANT USAGE ON SCHEMA public TO grafana_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA public TO grafana_reader;

\c bo_ats
GRANT USAGE ON SCHEMA bo_db_user TO grafana_reader;
GRANT SELECT ON ALL TABLES IN SCHEMA bo_db_user TO grafana_reader;

GRANT pg_read_all_stats TO grafana_reader;

Create function based on the basic SQL query:

To query the tenantID cross-database, use the dblink extension. The SELECT statement from dblink will return a table. Check if the extension has been created.

SELECT * FROM pg_extension WHERE extname = 'dblink';

CREATE EXTENSION dblink;

2. use maas_admin to login xservices_ems,and run below alert default command to let feature newly created table with select permission.

ALTER DEFAULT PRIVILEGES IN SCHEMA maas_admin GRANT SELECT ON TABLES TO grafana_reader;

3. Create functions in xservice_ems with postgres user. Here is the query definition.

For xservices_ems database: database_monitoring_functions_definition.txt

For xservices_rms database: database_monitoring_functions_definition_xservices_rms.txt

4. Create index for the query, this step is very important, please confirm this with expert.

create index "ops_transaction_context_<tenant_id>_timestamp" on transaction_context_<tenant_id> (transaction_timestamp);

create index ops_transaction_context_<tenant_id>_entity_type on transaction_context_<tenant_id> ((entity::jsonb ->> 'entity_type'));

This needs to be scripted and performed for every tenant in the farm.

Installation steps:

Download the helm chart prometheus-postgres-exporter.zip
Replace the <grafana_reader_password> and with actual value in the values_database_level.yaml,values_dead_tuple.yaml and values_instance_level.yaml.

Add User query in values_database_level.yaml file.

userQueries: |-
    nascm_transaction_context_queue_alltenants:
      query: "select tid,row_count from get_transaction_context_counts();"
      master: true
      cache_seconds: 1200
      metrics:
        - row_count:
            description: context queue count
            usage: GAUGE
        - tid:
            description: tenant id
            usage: LABEL
    nascm_transaction_context_queue_retries_alltenants:
      query: "select tid,row_count from get_transaction_context_queue_retries();"
      master: true
      cache_seconds: 1200
      metrics:
        - row_count:
            description: context queue count
            usage: GAUGE
        - tid:
            description: tenant id
            usage: LABEL
    etl_job_queue_alltenants:
      query: "select tid,etl_job_name,row_count from get_transaction_etl_job_queue();"
      master: true
      cache_seconds: 1200
      metrics:
        - row_count:
            description: job queue count
            usage: GAUGE
        - tid:
            description: tenant id
            usage: LABEL
        - etl_job_name:
            description: job name
            usage: LABEL
    etl_job_delay_time_alltenants:
      query: "select tid,etl_job_name,delay_time from get_transaction_etl_job_delay_time();"
      master: true
      cache_seconds: 1200
      metrics:
        - delay_time:
            description: delay time
            usage: GAUGE
        - tid:
            description: tenant id
            usage: LABEL
        - etl_job_name:
            description: job name
            usage: LABEL

run helm install command to deploy postgres exporter

helm install prometheus-postgres-exporter-instance-level. -f values_instance_level.yaml --namespace=monitoring

helm install prometheus-postgres-exporter-database-level. -f values_database_level.yaml --namespace=monitoring

helm install prometheus-postgres-exporter-for-dtuple. -f values_dead_tuple.yaml --namespace=monitoring

Configure Grafana to show metrics:

The metrics name is consist of _<metrics_name>. eg:

Import Dashboard to Grafana:

Download the PostgresSQL RDS Monitoring Dashboard and import to Grafana.

Related pages

Page: ESM Cloud Farm Version Tracking
Page: How to get an Opentext Confluence account
Page: ITOM APM AppPluse Cloud Farm Information
Page: ITOM Cloud Service Ops Doc Management Process
Page: ITOM ESM Cloud Service Catalog
Page: ITOM OpsB NOM Cloud Service Catalog
Page: OpsB and NOM Cloud Deployments Version Tracking

Attachments:

image-2025-5-26_14-53-21.png (image/png)
database_monitoring_functions_definition.txt (text/plain)
image-2025-5-26_15-15-4.png (image/png)
prometheus-postgres-exporter.zip (application/zip)
database_monitoring_functions_definition.txt (text/plain)
database_monitoring_functions_definition.txt (text/plain)
7. PostgreSQL RDS Monitoring-1751508520632.json (application/json)
7. PostgreSQL RDS Monitoring-1751508520632.json (application/json)
database_monitoring_functions_definition_xservices_rms.txt (text/plain)
database_monitoring_functions_definition_xservices_rms.txt (text/plain)

7.4 KiB Raw Blame History

How-to-deploy-postgres-exporter-on-SaaS-to-monitor-postgres-custom-query_704971984

Attachments:

7.4 KiB

Raw Blame History