Configuring Logging in Elastic Runtime

Page last updated:

This topic describes the types of logs that Pivotal Cloud Foundry (PCF) generates. It also explains how to forward system logs to an external aggregator service, how to scale Loggregator component VMs to keep up with app log volume, and how to manage app traffic logging.

System Logs, App Logs, App Traffic Logs

Pivotal Cloud Foundry (PCF) generates two types of logs, system logs from PCF components and app logs from hosted apps, as differentiated in the table below:

Log Type Originate from Follow format Stream from Can to stream out to (configurable) Visible to
System Logs Platform components Syslog standard rsyslog agent Component syslog drain Operators
App Logs Hosted apps Format is up to the developer Firehose1 External data platform, optionally via nozzles Developers and Operators
(Optional) With Scalable Syslog Adapter
Converted to syslog standard Scalable Syslog Adapter External syslog drain

1The Loggregator Firehose also streams component metrics.

App traffic logs are system logs. When app containers communicate, or attempt to communicate, their host cells generate app traffic logs. App traffic logs are system logs, not app logs. These logs come from host cells, not apps, and they carry no information from within the app. App traffic logs only show app communication behavior, as detected from outside by the host cell.

Enable Syslog Forwarding

To forward system log messages to an external Reliable Event Logging Protocol (RELP) server, complete the following steps:

  1. From the Elastic Runtime tile, Select System Logging. Sys logging
  2. To include security events in your log stream, select the Enable Cloud Controller security event logging checkbox. This logs all API requests, including the endpoint, user, source IP, and request result, in the Common Event Format (CEF).
  3. Enter the IP address of your syslog server in External Syslog Aggregator Hostname and its port in External Syslog Aggregator Port. The default port for a syslog server is 514.

    Note: The host must be reachable from the Elastic Runtime network, accept TCP connections, and use the RELP protocol. Ensure your syslog server listens on external interfaces.

  4. Select an External Syslog Network Protocol to use when forwarding logs.
  5. For the Syslog Drain Buffer Size, enter the number of messages the Doppler server can hold from Metron agents before the server starts to drop them. See the Loggregator Guide for Cloud Foundry Operators topic for more details.
  6. Click Save.

Scale Loggregator

Apps constantly generate app logs and PCF platform components constantly generate component metrics. The Loggregator system combines these data streams and handles them as follows. See Overview of the Loggregator System for more information.

  • The Metron agent running on each component or application VM collects and sends this data out to Doppler components.
  • Doppler components temporarily buffer the data before periodically forwarding it to the Traffic Controller. When the log and metrics data input to a Doppler exceeds its buffer size for a given interval, data can be lost.
  • The Traffic Controller serves the aggregated data stream through the Firehose WebSocket endpoint.

Follow the instructions below to scale the Loggregator system. For guidance on monitoring and capacity planning, see Monitoring Pivotal Cloud Foundry.

Add Component VM Instances

  1. From the Elastic Runtime tile, select Resource Config.
  2. Increase the number in the Instances column of the component you want to scale. You can add instances for the following Loggregator components:
    • Loggregator Traffic Controller

      Note: The Reverse Log Proxy (RLP) BOSH job is colocated with the Traffic Controller VM. If you want to scale Loggregator to handle more logs for syslog drains, you can add instances of the Traffic Controller.

    • Syslog Adapter
    • Doppler Server Loggregator vms
  3. Click Save.
  4. Click Apply Changes.

App Traffic Logging

App traffic logging generates logs when app containers communicate with each other directly, or attempt to communicate, as allowed by container-to-container networking (C2C) policies and Application Security Groups (ASGs).

App traffic logging lets network security teams audit C2C traffic, by seeing allowed and denied packets, without needing access to the Cloud Controller or the apps themselves.

Enable App Traffic Logging

To enable app traffic logging:

  1. From Ops Manager, navigate to the Pivotal Elastic Runtime tile > Networking pane.

  2. Under Log traffic for all accepted/denied application packets, select Enable (will increase log volume) or Disable to enable or disable app traffic logging.

Log app traffic enable

App Logging Behavior

App traffic logging generates log messages as follows:

  • TCP traffic - Logs the first packet of every new TCP connection.
  • UDP traffic - Logs UDP packets sent and received, up to a maximum per-second rate for each container. Set this rate limit in the UDP logging interval field (default: 100).
  • Packets denied - Logs packets blocked by either a container-specific networking policy or by Application Security Group (ASG) rules applied across the space, org, or deployment. Logs packet denials up to a maximum per-second rate for each container, set in the Denied logging interval field (default: 1).

App Traffic Log Format and Contents

App traffic logs are formatted as described in the cf-networking-release Traffic logging documentation, following the iptables-logger format but without line breaks. For example, the first part of an app traffic log line looks like: {"timestamp": "1500924070.182554722", "source": "cfnetworking.iptables", "message": "cfnetworking.iptables.ingress-allowed", "log_level": 1, "data": { "destination": { "container_id": "d5978989-1401-49ff-46cd-33e5","app_guid": "bc6f229d-5e4a-4c41-a63f-e8795496c283",.

Each log message includes the following:

  • Timestamp
  • The GUID for the source or destination app that sent or was designated to receive the packet
  • The protocol of the communication, TCP or UDP
  • GUIDs for the container, space, and org running the source or destination app
  • IP addresses and ports for both source and destination apps
  • A message field recording whether the packet was allowed or denied, with one of the following four possibilities:
    • ASG allowed packet to exit source app container
    • C2C policy allowed packet to enter destination app container
    • ASG prevented packet from exiting source app container
    • C2C policy prevented packet from entering destination app container
  • Additional information described in the cf-networking-release.

Denied Packet Causes

You can determine whether a denied-packet log resulted from a container networking policy or an ASG rule as follows:

  • Container networking policy: Log message string includes ingress-denied and packet direction is ingress.

  • ASG rule: Log message string includes egress-denied and packet direction is egress.

Global vs. ASG-Level App Traffic Logging

PCF supports two mechanisms for enabling app traffic logging. Setting Log traffic to Enable in Ops Manager enables app traffic logging globally for all ASGs and container policies. Setting the log property of an ASG to true enables app traffic logging at the individual ASG level.

Because these two mechanisms operate independently, PCF generates duplicate logs when app traffic logging is enabled globally and an ASG’s log property is set to true. To avoid duplicate logs, Pivotal recommends setting the log property to false for all ASGs, or leaving it out entirely, when app traffic logging is enabled globally.

To focus on specific ASGs for log analysis, Pivotal recommends enabling app traffic logs globally and using a logging platform to filter traffic logs by ASG, rather than setting log at the individual ASG level.

Create a pull request or raise an issue on the source for this page in GitHub