Data Retention

Primentra deletes old log and workflow records for you. All five retention periods live on one tab.

Access: Settings → General Settings → Retention

The Retention tab with the audit and approval retention cards(click to enlarge)

The five managed tables

Card	Table	Setting	Default	What is deleted
Audit Log Retention	`AuditLog`	`auditRetentionDays`	90 days	Change history: who changed what, when, and the old and new values
Approval Data Retention	`ApprovalRequests`	`approvalRetentionDays`	365 days	Resolved approval requests. Cascades to `ApprovalRows` and `ApprovalReviews`
Error Log Retention	`ErrorLog`	`errorLogRetentionDays`	30 days	Database error entries. The file `errorlog.txt` is not touched
Staging Data Retention	`StagingImport`	`stagingRetentionDays`	7 days	Processed staging records. Pending imports are never deleted
Scheduler Log Retention	`SchedulerLog`	`schedulerLogRetentionDays`	30 days	Scheduler dispatch events: fired, completed, skipped, zombie releases, errors

Every card carries the table name, a description, a number field from 1 to 9999 days, preset buttons, and an orange banner with the exact cutoff date. The banner updates live as you change the value.

Set a retention period

Open Settings → General Settings → Retention.
Type a number of days, or click a preset.
Read the orange banner to confirm the cutoff date.
Click Save settings.

Changes take effect at the next cleanup run. No restart is needed.

Approval data retention

Only requests in a terminal state are eligible:

approved — the changes were accepted and committed to live data
rejected — the changes were declined
sent_back — the request was returned to the submitter for revision
cancelled — the submitter withdrew the request

Pending requests are never deleted automatically. They still await a decision.

The cutoff is measured from the date the request was submitted, not from the date it was resolved.

This is the only card with a Never preset. Set it to 0 and cleanup is disabled — a blue panel reads *Approval records are kept indefinitely* instead of the orange cutoff banner.

How cleanup runs

A SQL Server Agent job runs usp_Maintenance_Cleanup every 60 seconds. Each run:

Reads the five retention values from the AppSettings table.
Calls the cleanup procedure for each table: usp_AuditLog_Cleanup, usp_Approval_Cleanup, usp_ErrorLog_Cleanup, usp_SchedulerLog_Cleanup and usp_Import_CleanupStaging.
Writes one SchedulerLog row per table — but only when records were actually deleted. Runs that delete nothing are silent.

The DELETE statements are cheap when there is nothing to purge, so the 60-second interval costs almost nothing.

Because the values live in AppSettings and not in a configuration file, the job always reads the current numbers, and the settings are included in a database backup.

Reading the cleanup log

A cleanup entry looks like this:

[Audit Log] Cleaned 42 records (retention: 90 days). Everything before 2025-12-02 was removed.

Filter for the cleanup event type in Settings → Logs, or query the table:

SELECT EventTime, Message
FROM   SchedulerLog
WHERE  EventType = 'cleanup'
ORDER BY EventTime DESC;

The job must exist

Without the SQL Server Agent job, none of these settings do anything — nothing is ever deleted.

Primentra checks for the job itself. Settings › General Settings › Database reports it after every setup run, and the Processing Schedule panel warns about it whenever a schedule is set to run by itself. You do not have to go looking.

To check by hand, connect as a sysadmin and run:

SELECT name, enabled FROM msdb.dbo.sysjobs WHERE name = 'Primentra_Scheduler';

No rows means it was not created. Run scripts\create-scheduler-agent-job.sql once as a sysadmin. Also confirm that the SQL Server Agent service is running and set to start automatically — the job cannot fire without it.

SQL Server Express has no SQL Server Agent at all. Use the Windows Task Scheduler alternative described in Staging Scheduler.

To purge once by hand, execute usp_Maintenance_Cleanup in SSMS at any time.