The DB is at max_connections=100 and several stack services hold persistent pools (several as the postgres superuser, idle for hours), so peaks hit "too many connections". The MCP is a minor contributor but easy to bound: - Dockerfile: uvicorn --workers 2 → 1. The MCP's connection budget is workers × MCP_POOL_MAX, so this caps it at 8 backends instead of 16. Scale via MCP_POOL_MAX, not workers, so the budget stays obvious. (Pairs with the minconn=0 lazy pool already on this branch: 0 connections held when idle.) - analytics_ro_role.sql: add idle_session_timeout=5min so the DB reaps the MCP's idle POOLED connections (idle_in_transaction never reaps them — they're idle outside a txn) and returns the slots. Safe because the server now discards + transparently retries a reaped connection instead of erroring. Note: the dominant fix is stack-wide (get the superuser app pools onto bounded, timed roles; consider PgBouncer; or raise max_connections) — out of this repo's scope but documented in the review. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
70 lines
4.4 KiB
SQL
70 lines
4.4 KiB
SQL
-- analytics_ro_role.sql — dedicated read-only LOGIN role for the analytics MCP server.
|
|
--
|
|
-- Sibling of dashboard_ro_role.sql, but for the decision & analytics team's MCP
|
|
-- server (analytics_mcp_rev.py) rather than the dashboard bridge. A separate role
|
|
-- keeps the two access paths independently revocable and lets us apply tighter,
|
|
-- human-ad-hoc-query guards (statement_timeout, idle-txn timeout) without touching
|
|
-- the dashboard bridge's credential.
|
|
--
|
|
-- Run as the postgres SUPERUSER (CREATE ROLE), NOT via run_migrations.py (which
|
|
-- connects as the app role and may lack CREATEROLE). Apply with
|
|
-- scripts/bootstrap_analytics_ro.sh, which supplies the password as the psql
|
|
-- variable :ro_pw from a host-only 0600 file — so no secret lives in this repo.
|
|
--
|
|
-- It grants exactly the read surface the MCP server needs:
|
|
-- * SELECT on reporting.* and tracksolid.* (tables + views)
|
|
-- * SELECT on the reporting.v_trips MATERIALIZED VIEW — matviews are NOT
|
|
-- covered by GRANT ... ON ALL TABLES, so it must be named explicitly
|
|
-- * EXECUTE on the reporting.fn_* functions (so analysts can SELECT reporting.fn_...)
|
|
-- * DEFAULT PRIVILEGES so future objects created by the migration role are
|
|
-- auto-readable (no re-grant when we add views)
|
|
-- Read-only: no INSERT/UPDATE/DELETE and not the matview owner, so analytics_ro
|
|
-- can never write or REFRESH. Idempotent -> safe to re-apply (also rotates pw).
|
|
|
|
\set ON_ERROR_STOP on
|
|
|
|
DO $role$
|
|
BEGIN
|
|
IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = 'analytics_ro') THEN
|
|
CREATE ROLE analytics_ro LOGIN NOSUPERUSER NOCREATEDB NOCREATEROLE;
|
|
END IF;
|
|
END $role$;
|
|
|
|
ALTER ROLE analytics_ro WITH LOGIN PASSWORD :'ro_pw';
|
|
|
|
GRANT CONNECT ON DATABASE tracksolid_db TO analytics_ro;
|
|
GRANT USAGE ON SCHEMA reporting, tracksolid, tickets, fuel TO analytics_ro;
|
|
|
|
GRANT SELECT ON ALL TABLES IN SCHEMA reporting TO analytics_ro; -- tables + views
|
|
GRANT SELECT ON ALL TABLES IN SCHEMA tracksolid TO analytics_ro; -- tables + views
|
|
GRANT SELECT ON ALL TABLES IN SCHEMA tickets TO analytics_ro; -- INC/CRQ tickets: tables + views
|
|
GRANT SELECT ON ALL TABLES IN SCHEMA fuel TO analytics_ro; -- fuel: tables + views
|
|
GRANT SELECT ON reporting.v_trips TO analytics_ro; -- MATERIALIZED VIEW (not in ALL TABLES)
|
|
GRANT EXECUTE ON ALL FUNCTIONS IN SCHEMA reporting TO analytics_ro;
|
|
GRANT EXECUTE ON ALL FUNCTIONS IN SCHEMA tickets TO analytics_ro;
|
|
GRANT EXECUTE ON ALL FUNCTIONS IN SCHEMA fuel TO analytics_ro;
|
|
|
|
-- "dynamic": future objects are auto-granted. reporting/tracksolid are created by the
|
|
-- migration role (tracksolid_owner); tickets/fuel are owned by postgres, so their
|
|
-- default-privilege grants must be keyed to postgres or new objects won't be readable.
|
|
-- NOTE: matviews are still never covered — a new matview needs its own explicit GRANT.
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE tracksolid_owner IN SCHEMA reporting GRANT SELECT ON TABLES TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE tracksolid_owner IN SCHEMA tracksolid GRANT SELECT ON TABLES TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE tracksolid_owner IN SCHEMA reporting GRANT EXECUTE ON FUNCTIONS TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE postgres IN SCHEMA tickets GRANT SELECT ON TABLES TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE postgres IN SCHEMA fuel GRANT SELECT ON TABLES TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE postgres IN SCHEMA tickets GRANT EXECUTE ON FUNCTIONS TO analytics_ro;
|
|
ALTER DEFAULT PRIVILEGES FOR ROLE postgres IN SCHEMA fuel GRANT EXECUTE ON FUNCTIONS TO analytics_ro;
|
|
|
|
-- Extra hardening over dashboard_ro: this role serves ad-hoc HUMAN queries via the
|
|
-- MCP server, so pin read-only at the role level and cap runaway work. These are
|
|
-- belt-and-braces alongside the read-only txn the server itself uses.
|
|
ALTER ROLE analytics_ro SET default_transaction_read_only = on;
|
|
ALTER ROLE analytics_ro SET statement_timeout = '30s';
|
|
ALTER ROLE analytics_ro SET idle_in_transaction_session_timeout = '60s';
|
|
-- Cap idle POOLED connections too (these sit idle but NOT in a transaction, so the
|
|
-- idle_in_transaction timeout never reaps them). On a shared 100-connection DB this
|
|
-- returns slots the MCP isn't using. Safe with the server's dead-connection handling:
|
|
-- a reaped pooled connection is discarded + transparently retried, not surfaced as an
|
|
-- error. (Requires PostgreSQL 14+.)
|
|
ALTER ROLE analytics_ro SET idle_session_timeout = '5min';
|