Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?


Centralized Logging

Juniper® Networks supports centralized logging using Cloud-Native Contrail Networking™ (CN2) Release 22.1 or later in a Kubernetes-orchestrated environment.

Benefits of Centralized Logging

  • The centralization of all platform logs eases troubleshooting. Centralization allows you (the administrator) to take a holistic view of events or outages extending across the many components within the deployment.

  • You have one portal, allowing you to monitor, view, filter, and search for events across all platform components from a single view.

Overview: Centralized Logging

Instead of browsing through individual log files, collected logs from all components of CN2 are available to you in a centralized location. The centralized location enables you to correlate the log files from multiple software components. For security, strict logging exists for all create, read, update, and delete (CRUD) actions. You perform these actions with individual access credentials so that you can identify individuals.

AWS OpenSearch Stack, an open source log collector and analyzer framework, provides out-of-box log collection and analysis functionality. The OpenSearch stack allows a single portal for analyzing logs from CN2. OpenSearch stack also analyzes logs from other software components and platforms deployed in the cluster. Examples include Linux OS logs, Kubernetes logs, and software components such as virtualized network functions (VNFs) and container network functions (CNFs).

OpenSearch Stack includes:

  • OpenSearch—Real-time and scalable search engine that allows for full-text and structured search as well as analytics. This search engine indexes and searches through large volumes of log data.
  • OpenSearch Dashboard—Allows you to explore your OpenSearch log data through a web interface and enables you to build dashboards and queries.
  • Fluentd—Logging agent that performs log collection, parsing, and distribution to other services such as OpenSearch.
  • Fluent Bit—Log processor and forwarder that collects data, such as metrics and logs, from different sources. High throughput with low CPU and memory usage. Fluent Bit is installed in every workload cluster.

The logging components are included and deployed in the optional telemetry node deployment. Installation commands are integrated in the telemetry installation.

Logs, Events, and Flows with Fluentd

Fluentd collects logs, events, and flows running on each CN2 node. Fluentd is the logging agent that performs log collection, parsing, and distribution to other services such as OpenSearch.

Figure 1: Logs, Events, and Flows with Fluentd Logs, Events, and Flows with Fluentd


Logs are collected from log files or stdout/stderr data streams and directed to the OpenSearch library stack with cluster quorum. Each CN2 node (configuration, control, compute, Web UI, and telemetry node) runs Fluent Bit or Fluentd to collect logs. The logs are sent to multiple configured sinks, such as OpenSearch. Fluentd supports multiple output options to send collected logs.

  • Control and compute nodes generate unstructured and structured logs through the Sandesh library. The Contrail Networking Sandesh library generates structured JSON files.

  • Configuration, Web UI, and telemetry node components produce standard logs to files or to stdout/stderr, which are then sent to Fluentd or Fluent Bit.

Multiple Kubernetes clusters in any CN2 cluster or in multiple CN2 clusters can connect with a Fluentd/OpenSearch monitoring component.


The vRouter agent and control node produce events through Sandesh. The Sandesh library produces JSON structured data and sends those files to the configured options. Configured options are stdout, file, or TCP port (Fluentd). Fluentd is configured with multiple output options to send data either to OpenSearch or to the telemetry node’s gRPC server. The telemetry node keeps cache for the latest status events.


The vRouter agent produces flow data at regular configured intervals. Configuration options for flow data generation supported by the vRouter agent are syslogs, JSON structure, and the default Sandesh.