Simplified dynamic profiling data model and heartbeat document #76
Conversation
| duration integer NULL DEFAULT 60, | ||
| frequency integer NULL DEFAULT 11, | ||
| profiling_mode ProfilingMode NOT NULL DEFAULT 'cpu', | ||
| target_hostnames text[] NOT NULL, |
There was a problem hiding this comment.
this should be extended to container names, pod names, namespace name
|
|
||
|
|
||
| -- Additional Types for Profiling System | ||
| CREATE TYPE ProfilingMode AS ENUM ('cpu', 'allocation', 'none'); |
There was a problem hiding this comment.
we also need those de-normalized tables to help with look up to map between namespace , services, jobs, pod, containers, and host
|
@dkorlovs and @skamerintel, let's review this PR together. |
There was a problem hiding this comment.
Pull Request Overview
This PR implements the database schema and documentation for a heartbeat-based dynamic profiling control system. The new system enables remote start/stop control of gProfiler agents through API commands, replacing the previous continuous profiling model with on-demand profiling capabilities.
Key Changes:
- Added four core database tables (HostHeartbeats, ProfilingRequests, ProfilingCommands, ProfilingExecutions) with supporting enums to track agent status and profiling command lifecycle
- Created comprehensive documentation covering architecture, API endpoints, configuration, and troubleshooting guidance
Reviewed Changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| scripts/setup/postgres/gprofiler_recreate.sql | Adds database schema for heartbeat-based profiling control with four new tables, custom enum types, and essential indexes |
| heartbeat_doc/README_HEARTBEAT.md | Provides complete documentation of the heartbeat system including architecture diagrams, API examples, configuration options, and operational guidance |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| -- FUNCTIONS | ||
|
|
There was a problem hiding this comment.
This comment header appears misplaced or misleading. The FUNCTIONS section is separated from the new profiling tables by the existing function definitions that start at line 357. Consider either moving this comment to just before line 357 where calc_profiler_usage_history begins, or removing it if it's not serving a structural purpose for the new schema additions.
| -- FUNCTIONS | |
| -- FUNCTIONS |
| -- Profiling Executions Table (optional - for audit trail) | ||
| CREATE TABLE ProfilingExecutions ( | ||
| ID bigserial PRIMARY KEY, | ||
| command_id uuid NOT NULL, | ||
| hostname text NOT NULL, | ||
| profiling_request_id uuid NOT NULL CONSTRAINT "fk_profiling_execution_request" REFERENCES ProfilingRequests(request_id), |
There was a problem hiding this comment.
Foreign key constraint references ProfilingRequests(request_id) but ProfilingRequests uses ID as the primary key. The constraint should reference the primary key column ID, not the UUID column request_id. Change to: REFERENCES ProfilingRequests(ID) and add a separate index on request_id if UUID lookups are needed.
| -- Profiling Executions Table (optional - for audit trail) | |
| CREATE TABLE ProfilingExecutions ( | |
| ID bigserial PRIMARY KEY, | |
| command_id uuid NOT NULL, | |
| hostname text NOT NULL, | |
| profiling_request_id uuid NOT NULL CONSTRAINT "fk_profiling_execution_request" REFERENCES ProfilingRequests(request_id), | |
| -- Add index for UUID lookups on ProfilingRequests | |
| CREATE INDEX idx_profilingrequests_request_id ON ProfilingRequests(request_id); | |
| -- Profiling Executions Table (optional - for audit trail) | |
| CREATE TABLE ProfilingExecutions ( | |
| ID bigserial PRIMARY KEY, | |
| command_id uuid NOT NULL, | |
| hostname text NOT NULL, | |
| profiling_request_id bigint NOT NULL CONSTRAINT "fk_profiling_execution_request" REFERENCES ProfilingRequests(ID), |
| -- Check command execution history | ||
| SELECT pe.hostname, pr.request_type, pe.status, pe.execution_time | ||
| FROM ProfilingExecutions pe | ||
| JOIN ProfilingRequests pr ON pe.profiling_request_id = pr.ID |
There was a problem hiding this comment.
This JOIN condition is inconsistent with the foreign key definition in the schema. The schema's foreign key constraint references ProfilingRequests(request_id), but this query joins on pr.ID. This query will fail since profiling_request_id contains UUID values from the request_id column, not bigint IDs. The correct join should be: JOIN ProfilingRequests pr ON pe.profiling_request_id = pr.request_id
| JOIN ProfilingRequests pr ON pe.profiling_request_id = pr.ID | |
| JOIN ProfilingRequests pr ON pe.profiling_request_id = pr.request_id |
Add Heartbeat-Based Dynamic Profiling Control Data Model
Overview
This PR introduces the database schema for a new heartbeat-based profiling control system that enables remote, dynamic control of gProfiler agents through API commands. This feature allows operators to start and stop profiling sessions on demand without manual intervention on target hosts.
Problem Statement
Previously, gProfiler agents ran continuously with fixed configurations, making it difficult to:
Solution
This PR adds four new core tables and supporting types to enable a robust heartbeat-based command-and-control system
Note : This is simplified version of the dynamic profiling data model. A more robust and scalable will be added in near future