Outsourced ProgrammersPowered by StealthAgents

How often do read replica grants outlive a reporting ticket?

A protocol for dating each replica grant, comparing it with the ticket window, and reporting the accounts that remain after closure.

How often do read replica grants outlive a reporting ticket?
WORK_LANE / RESEARCH BRIEF / DELIVERYREVIEW QUEUEACCEPTANCEPASSowner review requiredEVIDENCE TRAILTICKET → BRANCH → REVIEW → CLIENT RELEASE
Table of contents
  1. Research context
  2. Practical application
  3. Operating controls
  4. Sources
  5. Frequently asked questions

Research context

This protocol measures how often read replica grants outlive the reporting ticket they serve. The unit is one grant for one account on one replica tied to one ticket. An outliving grant means the account could still connect after the ticket closed. The study describes access closure, not whether the reporter used the data appropriately.

Define the replica endpoints and the ticket system in scope before collection. Record the database host, the role, and the review window. If the ticket process changes during the window, split the sample at the change date. A new process can change the closure evidence and the result.

Define a grant by the account, the replica, the role, and the ticket. Record the grant date, the expected close date, and the actual revoke time. A grant with no ticket is reported separately. Test the definition on the first five accounts.

Observe the database roles and the identity provider rather than the ticket alone. Look for an account that still authenticates after its ticket closed. Record the discovery source for each outliving grant. The study should report grants that the ticket system never knew about.

Check the query log for each grant to confirm that the account was used during the ticket window. A grant that was never used is a different finding from one that was used heavily. Record the last query time and the tables touched. Report the grants with no usable log.

Practical application

Check whether the account could reach the primary database or write to any table. A read-only grant that can reach the primary is a scope finding. Record the host, the role, and the tested permission. Report it separately from an account that stayed open too long.

Report the number of grants, the number still open past the close date, and the number with no ticket. Show the counts by team and by month. Present the grants with no revoke record and the grants with no log separately. Use the same denominator for every rate.

Separate a stale grant from a shared account. A stale grant has a named owner and an old ticket, while a shared account serves several people with no single owner. Report the two categories separately because the fixes differ. Record the evidence beside each finding.

Inspect the grants around a project handover or a team departure. A cluster may mean the closure step depends on a person who has moved on. Compare the pattern with the personnel record. A single stale grant can be an oversight, while a cluster is a process gap.

Do not treat a revoked grant as proof that the access ended. The account may still exist in the identity provider and regain access through another role. Test the login after revocation. Report what the test showed.

Operating controls

The record cannot show whether the account was used for an unapproved purpose or whether data was exported. It can show only whether the login still worked after the ticket closed. An outliving grant is not proof of misuse, and a closed grant is not proof of good oversight. The protocol also cannot see an account created outside the ticket process.

A useful additional view is the number of days between ticket close and revoke. A grant revoked the same day is a different control from one revoked a quarter later. Record both dates for each grant. That field turns a yes-or-no count into a review of closure speed.

Give the result to the client data owner and the offshore delivery lead with one observable change. The change might be a revoke-on-close step, a monthly grant review, a named owner per account, or a database and identity check. Record the owner, the effective date, and the measure that will show whether the change worked. Repeat the protocol with the same definitions after the change.

Close the review with one change and a date to re-measure. The change should address the largest outliving or shared-account group, not the easiest account to close. Record the expected effect and the evidence that will prove it. A grant review that is never re-measured is a cleanup, not a control.

Sources

  1. PostgreSQL hot standby
  2. NIST SP 800-53 security and privacy controls
  3. NIST SP 800-92 log management
  4. CISA Secure by Design

Frequently asked questions

What should happen first?

Begin with a bounded ticket, approved access, explicit acceptance evidence, and a named reviewer.

Who approves production changes?

The client's technical owner keeps final merge and release authority.

Related research

How often do TLS certificates expire inside an outsourced maintenance lane?

Build a programming lane your technical owner can review.

Bring the stack, backlog, acceptance checks, and release boundaries. We will help scope the role around the work.

Discuss your programming needs