Nebula Logger Purging And Retention

Log Retention and Purging in Nebula Logger

The Retention Model

Every Log__c record has a LogRetentionDate__c. When that date passes, the log becomes eligible for purging by LogBatchPurger. The retention date is set at insert time on Log__c from a combination of:

• LoggerSettings__c.DefaultNumberOfDaysToRetainLogs__c on the effective hierarchy record.
• Optionally, LoggerScenarioRule__mdt.NumberOfDaysToRetainLogs__c when the log's scenario matches a scenario rule (scenario-specific overrides win over the hierarchy default).

If no retention days are configured, LogRetentionDate__c is left null and the log is never auto-purged.

LoggerSettings__c Retention Fields

The retention side of LoggerSettings__c is a hierarchy custom setting - values cascade org default -> profile -> user. The relevant fields:

Purge Actions

Log__c.LogPurgeAction__c (populated from LoggerSettings__c.DefaultLogPurgeAction__c at insert time) tells LogBatchPurger what to do when a log ages out:

• Delete (default): The batch job hard-deletes the Log__c and its child LogEntry__c records. They do not sit in the Recycle Bin.
• Plugin-provided actions: Some Nebula Logger plugins add their own purge actions. For example, the big-object-archiving plugin copies logs into a big object before deletion so the data remains queryable long-term. When such a plugin is installed, its purge action becomes a valid value for LogPurgeAction__c.

Scheduling the Purge Job

LogBatchPurger does not run automatically. Schedule it via LogBatchPurgeScheduler:

// Run daily at 2am - adjust the cron expression for your org
String cronExpression = '0 0 2 * * ?';
System.schedule('Nebula Logger - Daily Purge', cronExpression, new LogBatchPurgeScheduler());

LogBatchPurgeScheduler has a no-arg constructor for default batch size and a LogBatchPurgeScheduler(Integer batchSize) constructor for tuning throughput.

Without a scheduled job, LogRetentionDate__c still gets set correctly on new logs, but nothing ever purges them. An unscheduled purger is the most common cause of "we set retention to 30 days but logs from six months ago are still there."

Per-Scenario Retention

To keep certain business processes' logs longer (or shorter) than the default, create a LoggerScenarioRule__mdt record whose Scenario__c matches the value passed to Logger.setScenario(...). Set:

• IsEnabled__c = true (otherwise the rule is inert).
• IsLogRetentionOverrideEnabled__c = true (this is the gate that actually enables the retention override - a NumberOfDaysToRetainLogs__c value with this flag off does nothing).
• NumberOfDaysToRetainLogs__c = the target retention in days for logs matching this scenario.

Use scenario rules for regulated processes ("keep financial-audit logs for 7 years") or for high-volume test scenarios ("purge integration-smoke-test logs after 24 hours") rather than trying to tune the hierarchy globally.

Per-User / Per-Profile Retention

Because LoggerSettings__c is a hierarchy custom setting, retention can be tuned per user or per profile. Common patterns:

• Debug users retain longer: Give developer profiles a longer DefaultNumberOfDaysToRetainLogs__c so investigation work isn't lost.
• High-volume integration users retain shorter: Give integration user profiles a shorter retention so their (much larger) log volume doesn't dominate storage.

Set the profile-level value in LoggerSettings__c for those profiles; it inherits from org default for everyone else.

Verifying Retention Setup

Before assuming retention is working, verify these three things in the target org:

1. LoggerSettings__c at the org default level has a non-null DefaultNumberOfDaysToRetainLogs__c.
2. LogBatchPurgeScheduler is scheduled (check Setup -> Scheduled Jobs).
3. New Log__c records get a LogRetentionDate__c populated. Run a quick test log and query SELECT Id, LogRetentionDate__c FROM Log__c ORDER BY CreatedDate DESC LIMIT 1.

If any of those three is missing, retention is not actually in effect.

Storage Considerations

• Log__c and LogEntry__c count against custom object storage. LogEntry__c is typically the volume driver (many entries per log).
• Higher LoggerSettings__c.LoggingLevel__c values (FINE, FINER, FINEST) produce dramatically more entries. Combine environment-appropriate logging levels with retention windows to control storage growth - see nebula-logger-best-practices.
• If storage becomes a real constraint, either shorten retention, raise logging level thresholds in production, or install the big-object-archiving plugin to offload aged data to a big object.

Related Skills

• nebula-logger-install - Initial LoggerSettings__c configuration.
• nebula-logger-best-practices - Environment-aware logging levels that pair with retention.
• nebula-logger-plugin-development - Building a custom purge-action plugin (e.g. archive to an external system).