Altoros Heartbeat for PCF
This documentation describes the Altoros Heartbeat for Pivotal Cloud Foundry (PCF) tile. Installing this tile creates a service that enables operators and app developers to track all system and application metrics from their PCF installation in one place.
Altoros Heartbeat for PCF is a multi-tenant monitoring service that uses Cloud Foundry’s User Account and Authentication (UAA) server to grant access to apps and resources based on the user roles and rights within a PCF deployment. The service pulls, stores, visualizes, and sends alerts on the metrics emitted by VMs, PCF components, services, and apps running on PCF.
The tile comes with pre-built integration with the following services available as PCF tiles:
This integration enables operators to monitor these services without any additional configuration.
The tile comes with pre-built integration with the following externally deployed services:
Altoros Heartbeat for PCF requires additional configuration to monitor these services.
Altoros Heartbeat for PCF includes the following key features:
- PCF monitoring: Collects metrics emitted by all components of PCF
- IaaS-level monitoring: Collects metrics from CPU, memory usage, IO, network, VM states, and more, regardless of the underlying infrastructure
- Third-party services monitoring: Provides the ability to collect metrics from MySQL, PostgreSQL, Redis, and RabbitMQ
- Application monitoring: Gathers standard metrics (CPU, memory, or disk usage), arbitrary metrics (gauges, counters, timings, sets), and HTTP-specific metrics (requests per second, status codes, error rates, and more)
- SSO and multi-tenancy: Uses the Cloud Foundry UAA service to control access to apps and resources based on user roles and rights within the platform. Multi-tenancy is implemented according to Cloud Foundry UAA roles. Dedicated shares of the solution are deployed to different spaces to collect metrics from the apps and services running within.
- Alerts: Can be defined for each dashboard panel, and allow users to get immediate notifications though email, Slack, and other notification channels.
The following table provides version and version-support information about Altoros Heartbeat for PCF.
|Release date||October 25, 2017|
|Software component version||Grafana v4.4.1, collectd v5.7.2, StatsD v0.8.0, carbonapi 0.8.0, carbon-c-releay 3.1, carbonzipper 0.73.2 and go-carbon 0.10.1|
|Compatible Ops Manager version(s)||v1.10.x and v1.11.x|
|Compatible Elastic Runtime version(s)||v1.10.x and v1.11.x|
|IaaS support||AWS and OpenStack|
To complete installation of Altoros Heartbeat for PCF, you need the following:
- An AWS S3 bucket or WebDAV with credentials to store alert images from Grafana
- Alert notification endpoint (SMTP server for email notifications, Slack, Sensu) with credentials
- CPU, RAM, and disk resources for seven VMs
- Internet connectivity between the networks where the PCF deployment and Heartbeat are installed. Altoros recommends installing Heartbeat on the same network with PCF.
- TCP 2003 and 8126, and UDP 8125 ports open in the firewall to accept connections from PCF and services networks. Ports 2003 and 8125 are used to receive metrics. TCP 8126 is necessary for Load Balancer health checks.
Altoros Heartbeat for PCF has the following requirements:
- Elastic Runtime v1.10.x, v1.11.x
- Ops Manager v1.10.x, v1.11.x
- Installation of the collectd add-on for PCF to enable PCF component VMs and the VMs of any other BOSH-deployed services expose component metrics to Heartbeat. For more information, see Installing collectd Add-on for PCF.
- (Optional) If you want to enable collection of JMX metrics from Java apps, you will need a custom Java buildpack for Altoros Heartbeat for PCF. For more information, see JMXTrans Agent Integration.
- (Optional) If you want to enable collection of custom application metrics, you must integrate a StatsD library into the app’s code. For more information, see StatsD Integration.
The Altoros Heartbeat for PCF tile has the following limitations:
- Offline buildpack for collecting JMX metrics is unavailable. Only the online version of the buildpack from this GitHub repository can be used to collect JMX metrics from Java apps.
- Users are limited to using internal MariaDB to store internal Heartbeat data such as user sessions, mappings, dashboards, UI configs, etc.
- Downscaling with retained high availability is only possible to at least two back-end instances.
Users interested in Heartbeat can sign up for a free trial.