Rudder - General settings

General settings

Allowed networks

Configure the networks from which nodes are allowed to connect to the Rudder policy servers to get their updated configuration policy.

You can add as many networks as you want, the expected format is: NetworkIP/mask, for example "42.42.0.0/16".

Security

Certificate's validation

This setting is kept for compatibility and only used for 6.X nodes Before enabling this, you have to replace self-signed certificates created by Rudder on your policy servers by valid certificates and to enforce using an hostname matching these certificates in policy_server.dat for all agents.

Relay synchronization

Configure the method used to synchronize files between Relay servers and the Rudder server.

The classic method doesn't require any setup and use the standard protocol. However, it does not scale beyond 1000 nodes per relay.

The rsync method triggers rsync synchronization between each Relay and the Rudder server, for the selected resources (policies and/or shared files). It is more efficient, but you need to manually set up rsync on the Relay servers, and proceed with the SSH key exchange. Note that ressources not selected below won't be synchronized.

Finally, the manual method disable all synchronization of policies and shared files between the Relay servers and the Rudder server; you will need to manually set up a transfer method.

Reporting behavior

The following settings affect the interpretation given to some type of unexpected reports when computing compliance. This option will take effect when the next reports are received from a node or if you "clear caches" below.

Unbounded reports by elements when a variable is used

Rudder await exactly one compliance report for each configuration point of control. If it gets more than one, the leftover reports will be considered as unexpected. This is not what one want in the case where:

  • the point of control is parametrized by a variable,
  • that variable is multivalued.

For example, if you build a technique in editor with the generic method "Variable iterator" to define a list of packages and you use the resulting variable in "Package present" generic method for the "name" parameter, it is normal and expected to get several compliance reports, one for each configuration value.

That option allows to increase the number of expected reports to the number of configuration values and so it avoids to get and "unexpected" compliance in that case.

Unless it is more important for you to get "unexpected" compliance than the actual compliance of each configuration value, you should check that option.

Default settings for new nodes

Configure the default state and policy mode for nodes when they are accepted within Rudder.

State may be used to classify nodes, in search and groups, and some states have also an impact on the policies generation for the node.

You may read the node lifecycle documentation for more information.

Modified files backup

Every time Rudder modifies a file (by file editing or copying from a remote source), a backup is created on the agent under /var/rudder/modified-files/.

Logging

All nodes in Rudder send reports via syslog to this Rudder root server. These logs are stored in an SQL database in order to determine compliance information displayed in this web interface. However, it can be useful to also store this information in a plain text log file, for example for statistics or debugging purposes. The option below enables this.

Also, the full output from each agent run is stored in a file under /var/rudder/cfengine-community/outputs/. These files are automatically removed to save on disk space. You can configure the retention time (Time To Live) they are kept for here.

Audit logs

If enabled, prompt users to enter a message explaining the reason for each configuration change they make.
These messages will be stored in each Event log and as the commit message for the underlying git repository in

Compliance display

In Rules table, we display a graph for each Rule showing its activity (number of repaired reports).

Unfortunately, some browsers (especially Firefox) have trouble displaying them and make Rule pages almost unusable.

If you experience slow loading of Rules pages, you can disable this feature here.

In directive configuration page, we have the possibility to choose rules for the directive. The rule are presented in a summary table which look alike the one in rule page. For performance on aesthetic reason, you may want to hide compliance and recent changes columns on that table. The column will still be displayed on the rule page.
[messages]

Script evaluation in Directives

If enabled, all Directive fields can contain a JavaScript expression. These expressions are evaluated during promise generation, and can therefore provide unique values for each node. Read the script documentation for more information.

Policy generation system hooks

Development Policy Check

Rudder can check generated policies syntax after generation to ensure that only syntactically valid policies are deployed on node..

That check is especially useful when you are developing your own generic method or technique without the technique editor, and you want to catch possible bugs as early as possible.

This check can add some overhead to policy generation, so if you don't code techniques or generic methods, you can disable it to make policy generation quicker.

[messages]

Trigger immediate update of nodes when their configuration changes

When new policies are generated, you have to wait for the next scheduled run of a node to have it apply them.

Depending on your use case, you may prefer to reduce latency between a node configuration change and its application on the node. For that, Rudder can send an update notification at the end of a policy generation to all nodes whose configuration has changed. That notification will tell the corresponding nodes to start a run immediately - and thus, not honouring neither their agent run period nor the splaytime between nodes. This can be a problem on large installation where hundreds of nodes would start a run concurrently, using a lot of CPU resources and network at the same time.

So by default, we advice to always limit the notification to a maximum number of nodes. Setting that number to 0 has the same effect than disabling that hook.

You can also limit the number of notified nodes to a fraction of the total updated node so that you can see on a small number of nodes how a change works. This is especially interesting if your schedule period is long, like 30 min, which let you some time to revert a faulty change. Of course, that help does not replace a real rollout strategy to deploy policies, and you should always test critical configuration changes on a dedicated group of nodes before rolling out the change on other production nodes.

If the number of nodes given by the ratio is bigger than the maximun number of nodes configured in the previous parameter, then the configured maximun number of node is used.

Nodes are chosen at random, with the exception of the policy servers that will always be notified first.

The ratio of nodes to update is given in percent, rounded up, so that 1% of 10 nodes will be 1 node, and 90% of 8 nodes will be all the 8 nodes. Setting that ration to 0 is the same than disabling that hook, and setting it to 100

The notification use the "trigger remote agent run" feature and you will need to open network port accordingly to that (see documentation).

[messages]

Debug information

Launch the debug script (/opt/rudder/bin/rudder-debug-info) to get information about your setup.

The provided information are used to troubleshoot Rudder.

Data includes various commands about your install (package version ...), your system and useful logs

[error]

Usage survey

To help the Rudder team continue to improve this software day after day, we are running a survey to collect usage statistics.

These statistics are submitted anonymously, and include overall statistics about your instance of Rudder (number of Rules, Directives, Nodes, etc). No potentially-sensitive data is included (only stock Rudder-provided techniques are examined, no hostnames, etc). We highly value your privacy, as we do our own, so we will never share individual submissions (only globally compiled statistics).

If you want to check the information that is sent, just run /opt/rudder/bin/rudder-metrics-reporting -v on your Rudder server. This won't send any information without your consent.

This information is very valuable to the development team, as it helps us focus on the features that matter most and better understand what our users care about. Please consider participating in the survey!

Reload groups

Groups in Rudder can be static (fixed list of nodes) or dynamic (the list of nodes is built from a search query).

To take into account new nodes and changes to their inventory, dynamic groups must be reloaded regularly.

Currently, Rudder will automatically do this reload every 5 minutes (see /opt/rudder/etc/rudder-web.properties).

Reload techniques

Techniques in Rudder are read from the filesystem (in /var/rudder/configuration-repository/techniques).

To take into account new Techniques and changes, the Technique library must be updated regularly.

Currently, Rudder will automatically do this update every 5 minutes (see /opt/rudder/etc/rudder-web.properties).