cloud-sql-postgres-monitor

name	cloud-sql-postgres-monitor
description	Use these skills when you need to troubleshoot performance bottlenecks, analyze query execution plans, identify resource-heavy processes, and monitor system-level PromQL metrics.

Usage

All scripts can be executed using Node.js. Replace <param_name> and <param_value> with actual values.

Bash: node <skill_dir>/scripts/<script_name>.js '{"<param_name>": "<param_value>"}'

PowerShell: node <skill_dir>/scripts/<script_name>.js '{\"<param_name>\": \"<param_value>\"}'

Note: The scripts automatically load the environment variables from various .env files. Do not ask the user to set vars unless skill executions fails due to env var absence.

Scripts

get_query_metrics

Fetches query level cloudmonitoring data (timeseries metrics) for queries running in Postgres instance using a PromQL query. Take projectID and instanceID from the user for which the metrics timeseries data needs to be fetched. To use this tool, you must provide the Google Cloud projectId and a PromQL query.

Generate PromQL query for Postgres query metrics. Use the provided metrics and rules to construct queries, Get the labels like instance_id, query_hash from user intent. If query_hash is provided then use the per_query metrics. Query hash and query id are same.

Defaults:

1. Interval: Use a default interval of 5m for _over_time aggregation functions unless a different window is specified by the user.

PromQL Query Examples:

1. Basic Time Series: avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m])
2. Top K: topk(30, avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
3. Mean: avg(avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
4. Minimum: min(min_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
5. Maximum: max(max_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
6. Sum: sum(avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
7. Count streams: count(avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))
8. Percentile with groupby on resource_id, database: quantile by ("resource_id","database")(0.99,avg_over_time({"__name__"="cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time","monitored_resource"="cloudsql_instance_database","project_id"="my-projectId","resource_id"="my-projectId:my-instanceId"}[5m]))

Available Metrics List: metricname. description. monitored resource. labels. resource_id label format is project_id:instance_id which is actually instance id only. aggregate is the aggregated values for all query stats, Use aggregate metrics if query id is not provided. For perquery metrics do not fetch querystring unless specified by user specifically. Have the aggregation on query hash to avoid fetching the querystring. Do not use latency metrics for anything.

1. cloudsql.googleapis.com/database/postgresql/insights/aggregate/latencies: Aggregated query latency distribution. cloudsql_instance_database. user, client_addr, project_id, resource_id.
2. cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time: Accumulated aggregated query execution time since the last sample. cloudsql_instance_database. user, client_addr, project_id, resource_id.
3. cloudsql.googleapis.com/database/postgresql/insights/aggregate/io_time: Accumulated aggregated IO time since the last sample. cloudsql_instance_database. user, client_addr, io_type, project_id, resource_id.
4. cloudsql.googleapis.com/database/postgresql/insights/aggregate/lock_time: Accumulated aggregated lock wait time since the last sample. cloudsql_instance_database. user, client_addr, lock_type, project_id, resource_id.
5. cloudsql.googleapis.com/database/postgresql/insights/aggregate/row_count: Aggregated number of retrieved or affected rows since the last sample. cloudsql_instance_database. user, client_addr, project_id, resource_id.
6. cloudsql.googleapis.com/database/postgresql/insights/aggregate/shared_blk_access_count: Aggregated shared blocks accessed by statement execution. cloudsql_instance_database. user, client_addr, access_type, project_id, resource_id.
7. cloudsql.googleapis.com/database/postgresql/insights/perquery/latencies: Per query latency distribution. cloudsql_instance_database. user, client_addr, querystring, query_hash, project_id, resource_id.

Parameters

Name	Type	Description	Required	Default
projectId	string	The Id of the Google Cloud project.	Yes
query	string	The promql query to execute.	Yes

---

get_query_plan

Generate a PostgreSQL EXPLAIN plan in JSON format for a single SQL statement—without executing it. This returns the optimizer's estimated plan, costs, and rows (no ANALYZE, no extra options). Use in production safely for plan inspection, regression checks, and query tuning workflows.

Parameters

Name	Type	Description	Required	Default
query	string	The SQL statement for which you want to generate plan (omit the EXPLAIN keyword).	Yes

---

get_system_metrics

Fetches system level cloudmonitoring data (timeseries metrics) for a Postgres instance using a PromQL query. Take projectId and instanceId from the user for which the metrics timeseries data needs to be fetched. To use this tool, you must provide the Google Cloud projectId and a PromQL query.

Generate PromQL query for Postgres system metrics. Use the provided metrics and rules to construct queries, Get the labels like instance_id from user intent.

Defaults:

1. Interval: Use a default interval of 5m for _over_time aggregation functions unless a different window is specified by the user.

PromQL Query Examples:

1. Basic Time Series: avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m])
2. Top K: topk(30, avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
3. Mean: avg(avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
4. Minimum: min(min_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
5. Maximum: max(max_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
6. Sum: sum(avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
7. Count streams: count(avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))
8. Percentile with groupby on database_id: quantile by ("database_id")(0.99,avg_over_time({"__name__"="cloudsql.googleapis.com/database/cpu/utilization","monitored_resource"="cloudsql_database","project_id"="my-projectId","database_id"="my-projectId:my-instanceId"}[5m]))

Available Metrics List: metricname. description. monitored resource. labels. database_id is actually the instance id and the format is project_id:instance_id.

1. cloudsql.googleapis.com/database/postgresql/new_connection_count: Count of new connections added to the postgres instance. cloudsql_database. database, project_id, database_id.
2. cloudsql.googleapis.com/database/postgresql/backends_in_wait: Number of backends in wait in postgres instance. cloudsql_database. backend_type, wait_event, wait_event_type, project_id, database_id.
3. cloudsql.googleapis.com/database/postgresql/transaction_count: Delta count of number of transactions. cloudsql_database. database, transaction_type, project_id, database_id.
4. cloudsql.googleapis.com/database/memory/components: Memory stats components in percentage as usage, cache and free memory for the database. cloudsql_database. component, project_id, database_id.
5. cloudsql.googleapis.com/database/postgresql/external_sync/max_replica_byte_lag: Replication lag in bytes for Postgres External Server (ES) replicas. Aggregated across all DBs on the replica. cloudsql_database. project_id, database_id.
6. cloudsql.googleapis.com/database/cpu/utilization: Current CPU utilization represented as a percentage of the reserved CPU that is currently in use. Values are typically numbers between 0.0 and 1.0 (but might exceed 1.0). Charts display the values as a percentage between 0% and 100% (or more). cloudsql_database. project_id, database_id.
7. cloudsql.googleapis.com/database/disk/bytes_used_by_data_type: Data utilization in bytes. cloudsql_database. data_type, project_id, database_id.
8. cloudsql.googleapis.com/database/disk/read_ops_count: Delta count of data disk read IO operations. cloudsql_database. project_id, database_id.
9. cloudsql.googleapis.com/database/disk/write_ops_count: Delta count of data disk write IO operations. . , .

Parameters

Name	Type	Description	Required	Default
projectId	string	The Id of the Google Cloud project.	Yes
query	string	The promql query to execute.	Yes

---

list_active_queries

List the top N (default 50) currently running queries (state='active') from pg_stat_activity, ordered by longest-running first. Returns pid, user, database, application_name, client_addr, state, wait_event_type/wait_event, backend/xact/query start times, computed query_duration, and the SQL text.

Parameters

Name	Type	Description	Required	Default
min_duration	string	Optional: Only show queries running at least this long (e.g., '1 minute', '1 second', '2 seconds').	No	1 minute
exclude_application_names	string	Optional: A comma-separated list of application names to exclude from the query results. This is useful for filtering out queries from specific applications (e.g., 'psql', 'pgAdmin', 'DBeaver'). The match is case-sensitive. Whitespace around commas and names is automatically handled. If this parameter is omitted, no applications are excluded.	No	``
limit	integer	Optional: The maximum number of rows to return.	No	50

---

list_database_stats

Parameters

Name	Type	Description	Required	Default
database_name	string	Optional: A specific database name pattern to search for.	No	``
include_templates	boolean	Optional: Whether to include template databases in the results.	No	false
database_owner	string	Optional: A specific database owner name pattern to search for.	No	``
default_tablespace	string	Optional: A specific default tablespace name pattern to search for.	No	``
order_by	string	Optional: The field to order the results by. Valid values are 'size' and 'commit'.	No	``
limit	integer	Optional: The maximum number of rows to return.	No	10

---

list_locks

Identifies all locks held by active processes showing the process ID, user, query text, and an aggregated list of all transactions and specific locks (relation, mode, grant status) associated with each process.

---

list_query_stats

Lists performance statistics for executed queries ordered by total time, filtering by database name pattern if provided. This tool requires the pg_stat_statements extension to be installed. The tool returns the database name, query text, execution count, timing metrics (total, min, max, mean), rows affected, and buffer cache I/O statistics (hits and reads).

Parameters

Name	Type	Description	Required	Default
database_name	string	Optional: The database name to list query stats for.	No	``
limit	integer	Optional: The maximum number of results to return. Defaults to 50.	No	50

---

long_running_transactions

Identifies and lists database transactions that exceed a specified time limit. For each of the long running transactions, the output contains the process id, database name, user name, application name, client address, state, connection age, transaction age, query age, last activity age, wait event type, wait event, and query string.

Parameters

Name	Type	Description	Required	Default
min_duration	string	Optional: Only show transactions running at least this long (e.g., '1 minute', '15 minutes', '30 seconds').	No	5 minutes
limit	integer	Optional: The maximum number of long-running transactions to return. Defaults to 20.	No	20

---