DKDKCISSP
← Cybersecurity on the GroundCHALLENGE

The Setting Change That Broke the Audit Proof

By Deepak Kumar, CISSP

The problem: The security operations team (the SOC) is running out of disk space on the firewall log server. An engineer changes the setting from keeping logs for 12 months to keeping them for 90 days. Nobody tells risk and compliance, whose audit proof says “firewall logs are kept for a year.” Months later the auditor asks for logs from eight months ago. They are gone.

Words used here: the SOC is the team that watches alerts and deals with attacks. A log is the record a system keeps of what happened. Retention is how long a record is kept before it is deleted. Evidence (proof) is something that shows a check really happened.

Think of a bus timetable

Every morning Priya catches the 8:10 bus. The bus company changes its timetable to save fuel and puts up no notice. Priya stands at the stop and watches an empty road. The company did nothing silly: the new timetable works fine for the company. Priya was simply never told.

The SOC’s change is the same. It makes sense for the SOC. The people who rely on the old setting were never told.

The bus timetable nobody posted
Bus stop: the 8:10 no longer runs
The new timetable saves fuel.
Bus company
Nobody told me.
Priya
The change was sensible. The people relying on it were not told.

What it looks like at work

The firewall log server is almost full. A full disk would stop logging altogether, so a SOC engineer sets retention to 90 days. The ticket is closed as done, and logs older than 90 days are deleted overnight.

Compliance does not know. Its yearly control sheet still says “firewall logs kept for 12 months,” copied from last year. At the audit, the auditor asks for logs from eight months ago. Nobody can produce them. The control is marked as failed.

How a quiet setting change breaks the audit proof
  1. 11. Disk nearly fullLogging could stop altogether
  2. 22. Setting changed12 months cut to 90 days
  3. 33. Nobody toldTicket closed, compliance unaware
  4. 44. Audit asksLogs from eight months ago
  5. 55. Control failsThe logs are gone for good
The setting looked routine. The audit depended on it.

Both teams have a fair point. The SOC has a real problem: a full disk stops all logging, which is worse than keeping less. Compliance has a real need: an auditor accepts proof only for the period the policy states. Nobody was careless. The gap is that a setting the audit depends on looks, to the SOC, like any other setting.

You will notice it when:

  • A setting is changed to fix a daily problem and nobody outside the team hears about it.
  • Compliance’s control sheet says one thing and the tool says another.
  • Closing a ticket as “done” is the only record of a change.
  • Evidence that existed last year cannot be found this year.
  • The auditor tests a control and finds it behaves differently from the policy.

What exactly needs fixing

The first fix is not the disk. It is making sure that a setting the audit relies on cannot change in silence. This card shows the whole problem on one page:

A setting changed in silence
WHAT IS BROKEN

Firewall log retention was cut from 12 months to 90 days, and compliance was not told.

HIGH
WHERE IT IS

The firewall log server, and compliance’s yearly control sheet.

WHAT COULD GO WRONG

The audit control fails, and old logs that an investigation needs are gone for good.

WHAT IT COSTS US
  • Failed control
  • Lost old logs
  • Weeks of explaining
  • Auditor trusts proof less
THE FIX, IN THREE MOVES
  1. 1
    1. Restore the settingBack to 12 months, or move old logs to cheaper storage
  2. 2
    2. List audit settingsRetention, log sources, backups, each with an owner
  3. 3
    3. Share a change logChanges to listed settings are posted two days ahead
WHO FIXES IT

Who fixes it: SOC lead and compliance lead

BY WHEN

By when: setting restored this week, list in 2 weeks

A blank copy of this card is in the Tools section.

Why it happens

  • The setting looks like routine housekeeping. Nobody marks which settings the audit relies on.
  • There is no shared place to announce changes. The ticket system tells IT, not compliance.
  • Compliance copies last year’s answer. The control sheet is updated at audit time, not when the tool changes.
  • The disk problem is urgent and the audit is months away. Urgent wins.
  • Nobody owns the link. The SOC owns the tool and compliance owns the proof, but nobody owns the step in between.

What happens if you ignore it

The control fails at the audit, and the cause is a setting nobody remembers changing.

  • The auditor writes a finding that a control “is not operating as described.”
  • Old logs that an investigation might need are gone for good.
  • The team spends weeks explaining and rebuilding what it can.
  • The same change can happen again with the next setting.

At a glance

QuestionAnswer
How likely is it to hurt us?Likely, wherever tools are tuned by one team and audited by another.
Worst case?An old attack cannot be investigated because the logs were deleted.
What does it cost?A failed control, weeks of explaining, and an auditor who trusts the proof less.
What if we wait?The next setting change breaks the next control.
What does the fix cost?One list, one shared change log, and a two-day warning habit.

How to fix it, step by step

Mark the settings, warn before changing them, then check afterwards.

  1. Restore the setting first. Put retention back to 12 months, or move old logs to cheaper storage so the disk is not the reason to cut it.
  2. Tell the auditor early. If logs are already lost, say so plainly and show the plan, before they find it.
  3. List the settings the audit relies on. Retention, log sources, backup times, review schedules. Give each one an owner.
  4. Start one shared change log. A simple page or sheet. Any change to a listed setting goes in it two days before it happens.
  5. Make compliance a reader of that log. They look at it weekly and update the control sheet when something changes.
  6. Add one line to the SOC’s change ticket. “Does this setting feed an audit control? Yes or no.”
  7. Check after every change. Compliance confirms the proof still shows what the policy says.
Mark, warn, then check
1. MarkList the settings the audit relies on.
2. WarnPost changes two days ahead in one log.
3. CheckConfirm the proof still matches the policy.
The warning step stops it happening again.

The picture shows the order: mark the settings, warn before changing, then check. The warning step is the one that stops this happening again.

Who to bring in: The SOC lead (owns the tool settings), the compliance lead (owns the list and the proof), the head of IT if the server is theirs, and the internal auditor (agrees what proof is accepted).

Tools and templates

Settings the audit depends on

SettingValue the audit expectsOwnerTold before a change
Firewall log retention12 monthsSOC leadCompliance lead
Admin login log retention12 monthsIT teamCompliance lead
Backup scheduleAs the policy statesIT teamCompliance lead

Blank finding card

FieldWrite here
What is brokenOne line, in plain words
Where it isThe system, team, or process
What could go wrongIn one sentence
What it costs usMoney, time, a failed audit, lost trust
The fixThe exact action
Who fixes itOne name or one team
By whenA date, not “soon”

Email to start the talk

Subject: A shared list of settings the audit depends on

Hi both. We found that the firewall log retention was changed from 12 months to 90 days, and compliance was not told. I would like us to keep one short list of settings the audit relies on, and a shared place where changes are posted two days ahead. Can we meet for 30 minutes this week? I am not looking for who is at fault. I am looking for a way to make sure it cannot happen quietly again.

Field tip

Add one question to every change ticket: “Does this setting feed an audit control?” It costs five seconds and catches most of these.

Words used in this chapter

  • SOC: the security operations team that watches alerts and deals with attacks.
  • Log: the record a system keeps of what happened.
  • Retention: how long a record is kept before it is deleted.
  • Control: a check or safeguard that must happen.
  • Evidence (proof): something that shows a check really happened.
  • Finding: a problem the auditor writes down.