PingDirectory

Replication best practices

This section covers the best practices for replication based on our observations in production environments.

Purging obsolete replicas

About this task

By default, the PingDirectory server retains replicas for every server that has been part of a given topology. Over time, this can lead to an accumulation of obsolete replicas in your topology, which can increase the complexity of replication and negatively impact reliability.

To purge these obsolete replicas by default, you can set the replication-purge-obsolete-replicas global configuration property to true.

If you set replication-purge-obsolete-replicas to true, you must make this change on all server replicas, including new replicas as they are added. Failure to do so could cause replicated servers to communicate an inconsistent state across the topology. An inconsistent state can lead to a replication-missing-changes alert and a server lockdown because of the missing replication changes.

Steps

  • To set replication-purge-obsolete-replicas to true, run dsconfig set-global-configuration-prop, as in the following example:

    Example:

    bin/dsconfig set-global-configuration-prop \
        --set replication-purge-obsolete-replicas:true

About the dsreplication command-line utility

The following points involve some security practices that apply to replication. For specific questions, contact your authorized support provider.

Developing Scripts

The dsreplication utility maintains the history of executed dsreplication commands with the full command-line arguments in the logs/tools/dsreplication.history file. Use the recorded commands to develop scripts to set up and configure replication.

Scripts invoking the dsreplication utility in non-interactive mode check the return code from the dsreplication process. A non-zero return code indicates a failure.

To exclude output messages from the dsreplication utility, use the --quiet option to suppress them.

By default, the utility fails if one or more warnings are issued during the command execution. To suppress warnings, use the --ignoreWarnings option.

This option is required when using dsreplication with non-fully-qualified host names, such as localhost. Otherwise dsreplication fails. However, you should avoiding using this flag in production environments.

Concurrent Use

With the exception of using the dsreplication utility with the status subcommand, the dsreplication subcommands cannot execute concurrently. The command-line utility locks the replication topology at one or more servers to prevent accidental configuration changes caused by multiple dsreplication subcommands running at the same time. For best practices, avoid making concurrent configuration changes.

Status

Using the dsreplication utility with the status subcommand requires the Replication Servers to provide monitoring information. This can lead to a delay before the output of dsreplication status displays. By default, dsreplication displays the status for all replicated domains with the exception of the special domains of the schema and the server registry.

Select a particular base distinguished name (DN) for dsreplication status if multiple base DNs are configured for replication.

Avoid invoking dsreplication status too often, such as more than once every 15 seconds or from multiple locations at the same time. Some of the information displayed by dsreplication status is based on monitor information that does not refresh every time the monitor is queried.

Do not use the status subcommand for collecting performance metrics. It is best to use the Periodic Stats Logger plugin for replication-related information.

Topology Tasks

The dsreplication tool allows specifying more than one server in a topology to act as the host for other servers for the enable and initialize actions. Use the --topologyFilePath option to specify a file with a series of hosts and ports in the topology.

For example, when the hosts file is used for an enable or initialize action, the servers in the file are tried sequentially until the new server is successfully enabled or added. The rest of the servers in the file are ignored. This ensures that a host server is always available. This file generates with the manage-topology export command.