Daily checks and routine maintenance
Documentation for Tutela v3.9.0.
Use this checklist to spot unhealthy services, unfinished reviews, missing usage, and backup failures. Record the affected item and who will resolve it; checking that a page opens is not enough.
Prerequisites
- Access to your deployment and permission to view the areas you are responsible for.
- A named deployment operator for infrastructure changes and an administrator for access or policy changes.
- A list of the clients your organization intends to connect, their owners, and the most recent successful backup and restore rehearsal.
Before handing the deployment to your team
- Complete Verify your installation for each connection you intend to use. A browser extension, API client and MCP client follow different paths; one passing test does not verify all three.
- Record the deployment URL and platform version, plus the installed extension and SDK versions. Confirm who can access the AWS stacks or Helm release when infrastructure repair is needed.
- Open Settings > Runtime setup and compare the configured clients with your rollout list. A discovered application is an inventory observation; confirm a recent inspected request before treating a client as connected to protection.
- Review Settings > Backups and complete a restore rehearsal in an isolated environment before relying on the archive.
Daily checks
| Check | Where to look | Expected condition | If it differs |
|---|---|---|---|
| Service availability | Settings > Support & diagnostics > Service health | Required dependencies respond; no unresolved unhealthy state | Record the service and time; follow Service unavailable. |
| Requests and findings | Overview, then Activity and Findings | Recent activity from the clients in use; findings have a reviewer | Check time filters and connection scope. Investigate unexpected gaps or high-severity findings. |
| Open investigations | Cases | Active investigations have an owner and a next action | Assign ownership and add the evidence or follow-up needed. |
| Pending decisions | Tools & MCP Servers > Approvals | Pending requests have an eligible reviewer and adequate context | Review the protected proposal; do not approve a request simply to clear the queue. |
| Usage and budgets | AI Spend | Reporting has a recent observation time for active sources; missing prices and imports are understood | Use AI usage and cost problems. |
| Scheduled backups | Settings > Backups | The scheduled archive has a recorded successful result | Inspect Backup History and operation status; use Backup and restore. |
Choose a time range that includes your organization's working hours. An idle weekend and a failed collection service can both produce no new activity; use source health and the last recorded event to distinguish them.
When activity information is missing
For administrators and deployment operators using v3.9, Activity collection needs review means Tutela recorded a problem receiving or processing activity. It does not prove that an action failed, that information was stolen, or that every connected application is affected.
- Check the selected period and ask the connection owner whether activity was expected. No new activity during an idle period is not, by itself, a failure.
- Open the source information and note the affected connection, check time and timezone. Distinguish the last successful receipt from when Tutela last checked the source. If a receipt time is unavailable, record that limitation; do not substitute the check time.
- Use Check service health when available, or open Settings > Support & diagnostics. Give the connection owner the source name and displayed problem. If your role cannot access these details, ask an administrator to review them.
- Continue reviewing available action records separately. A completed action can have an incomplete security check. Several findings may share one collection problem; track that repair once with its owner.
- After the owner repairs the connection, confirm that new expected activity is arriving and the source check has updated. Existing gaps remain unknown unless retained records actually fill them. Escalate an unchanged problem using the support handoff below.
Do not turn off protection policies to clear a collection warning. Do not replay a real action to manufacture a missing record, replace credentials without diagnosing the failed connection, or include prompts, secrets or customer files in a support report.
Verification status: this new-release procedure is not yet verified in a deployed environment. It remains subject to release acceptance; a successful documentation build does not establish that a connection was repaired.
Weekly and after a change
- Review Evidence > Audit records for unexpected changes to users, credentials, policies and integrations. Confirm that the owner recognizes each change, including changes made through administrative automation.
- Review new clients and retired connections with their owners. Check a synthetic request after changing a client, provider, policy or gateway route.
- Check storage growth and resource alerts with the deployment operator. Schedule capacity work before limits interrupt requests; use the relevant runbook.
- Review backup retention and the last restore rehearsal. Schedule a new rehearsal after changes that affect stored data or recovery, and at the interval agreed in your recovery plan.
- Record outstanding issues, owners and dates. Review the previous checklist before starting the next one.
Support handoff
For an unresolved problem, include the deployment address, platform and affected client versions, time and timezone, affected task, visible error, and request or Case ID. Explain what you tried and the result. Operators can include relevant stack events and health results after removing sensitive content.
Open Settings > Support & diagnostics to prepare a problem report. If Tutela is unreachable, contact support@h2htech.com. Do not send passwords, access credentials, cookies, authorization headers, or customer prompts and files.
Validate
- Every failed check has an owner and a recorded next action.
- Changes were followed by a test of the affected connection.
- Backup completion and restore readiness are recorded separately.
- Remaining uncertainty is documented rather than marked healthy.