Foglight 5.9.1 - High Availability Field Guide

Configuring Foglight Agent Manager for HA mode

The Foglight™ Agent Manager client fully supports communicating with a Management Server HA cluster. During Agent Manager installation, you can configure multiple HA URLs for the upstream Management Server. In addition, if you connect to the active Management Server in an HA cluster, the Agent Manager installer can automatically query for and configure the other HA peer URLs. This ensures that, if the active (primary) Management Server fails over to an HA peer (secondary server), the Agent Manager clients will connect to the new active peer seamlessly.

For more information about installing and configuring the Agent Manager, see the Foglight Agent Manager Installation Guide.

Upgrading cartridges in a High Availability environment

To upgrade cartridges in an HA environment, you only need to install the cartridge you want to upgrade on the primary server while both the primary and the secondary Management Servers are running. The cartridges are then copied over to the secondary server automatically.

The management log reports on the copying of the cartridges.

Troubleshooting FAQ

If Management Servers that are not part of a particular HA partition are configured to multicast using the same port on the same subnet as those within the HA partition, errors like the following appear in the logs for the Management Servers that are configured in this way:

ERROR [UDP mcast receiver] org.jgroups.protocols.UDP - message does not have a UDP header

or

ERROR [UDP mcast receiver] org.jgroups.protocols.UDP - discarded message from different group

If you see this type of message, change the cluster multicast port of the servers that are reporting the error. In addition, if you plan to have multiple High Availability (HA) partitions on the same subnet, configure a different cluster multicast port for each HA partition. See Configuring the cluster multicast port used in HA mode for instructions.

IMPORTANT: In both cases described above (resolving errors and planning to have multiple HA partitions on the same subnet), the primary server and secondary servers within each HA partition must be configured to use the same cluster multicast port.

Each Management Server in an HA cluster checks its own internal health.monitor.port to try and determine the server health. If this check fails, the server is automatically restarted. The HA Health Check Monitor does not use HTTPS, so when the Management Server is configured to use HTTPS/SSL, the port health check fails and the server is restarted.

To prevent this issue, configure the Health Check URL in the restart_monitor.config to use HTTP and port 8080.

By default, the URL is:
health.check.url=”https://localhost:8443/foglight-sl/HealthCheck”;

Specify HTTP as the transport protocol, and define the actual Management Server host name and port 8080. For example:
health.check.url=”http://hostname:8080/foglight-sl/HealthCheck”;

Ensure that the URL is updated on each server in the cluster.

Common configuration scenarios

This section outlines three common configuration scenarios for HA mode. It is intended only to illustrate possible configurations and processes and does not imply that Quest favors or recommends any of the platforms, technologies, or procedures described.

This information is intended for anyone taking the role of architect or installer for a Foglight™ 5.x project.