Change collision worked case: two approved changes, one outage, one rollback

Short answer

The records that explain this kind of failure are the ones that show two changes passing review separately, the failure they caused together, the decision about which one to roll back, and the fix to the process afterward: both change requests and their approvals, the conflict check result, the monitoring alert, the incident with its correlated recent changes, the emergency rollback approval, the restore time and the post-incident review. Each change record looks clean alone. The value is in the set.

Visual briefFull-size diagram
An overnight timeline from 01:00 to 03:00: change A and change B run in overlapping windows, an alert fires at 01:40, the case branches into an incident, a rollback approval at 02:05 and service restored at 02:20, then a post-incident review adds the missing dependency.
Illustrative collision for a fictional software and hosting company; times and roles are invented for the example.

This worked case belongs to the change collision recovery guide, which covers what a platform merger can leave behind. Here the job is narrower: follow one bad night from the first approval to the final fix, mark who decided each step and which record proves it, then turn that into signs an ITSM implementation partner can recognize at a client.

The case: a database patch and a load balancer change on the same night

Illustrative. A fictional 650-person software and hosting company runs its change process in one tool and its on-call response in a separate incident platform. On one Saturday night two changes were scheduled.

  • Change A patched a database cluster. A normal change, approved at the weekly change advisory board, planned 01:00 to 03:00.
  • Change B updated a load balancer configuration. A pre-approved standard change, planned from 01:20 the same night.

Each change was correct for the component it touched. The checkout service, however, relied on both the database cluster and the load balancer, and nobody had written that link into the inventory of components. That is the exception: two individually approved changes that conflict.

The night, step by step

StepTimeWhat happenedWho decidedRecord that proves it
1WednesdayChange A approved for Saturday 01:00 to 03:00Change advisory boardChange request, approval, CAB notes
2Pre-approvedChange B scheduled from 01:20Standard change template ownerChange request on a standard template
3Before the windowConflict check on both: no conflicts listedThe ITSM toolStored conflict result
401:40Checkout error rate crosses its alert thresholdMonitoring rulesAlert with timestamp
501:45Incident opened; change B shown as a recent change, correlated by timeOn-call engineerIncident with recent-change correlation
602:05Roll back change B; let change A finishChange managerEmergency rollback approval with reason
702:20Alert clears; incident resolvedOn-call engineerIncident resolution time
8Following weekReview finds the missing link and records itProblem managerPost-incident review; inventory relationship change

Reading each record: what it proves

The table lists the records. The point of a worked case is what each one shows that the others cannot.

  • Step 3, the empty conflict result. ServiceNow's conflict detection works from the change's configuration item and planned dates, and reports issues such as CI already scheduled, not in maintenance window, or blackout. Changes A and B sat on different CIs, so the check had nothing to flag. A stored "no conflicts" result proves the check ran and records the limit of what it could see.
  • Step 5, the correlation. In PagerDuty, an incident's Recent Changes section shows up to three changes correlated from the past 24 hours, with a reason such as time or related service. Related-service matches depend on configured service dependencies; here only time linked them. The record shows how the suspect was found, and why a better dependency map would have found it faster.
  • Step 5 also depends on step 2 being visible. Change events do not page anyone; they appear only as context. PagerDuty's ServiceNow integration can create those change events when a change request is opened. If the integration was off, step 5 would show nothing.
  • Step 6, the choice under pressure. Rolling back the database patch mid-window would have been riskier than undoing a configuration change. The change manager's approval note, with its reason, records a judgment no rule makes.
  • Step 8, the correction. The CMDB relationship added after the review is what stops the next collision. Its date matters: a dependency added after the incident should not be read as one that existed before.

Why this sequence is useful evidence

Records 3, 6 and 8 carry the judgment: what the review process could not see, which change to sacrifice, and what was changed so it would not recur. Public documentation describes how conflict detection and correlation work. It cannot show an experienced team deciding, at 02:05, which of two approved changes to undo.

For AI developers building IT operations agents, that ordered chain of decision, outcome and correction is the scarce part, as why AI agents need work data explains. Agents tend to fail exactly where the documented process runs out, a theme covered in why AI agents fail. The same problem, action and outcome shape appears in freight in logistics exceptions and their documented resolutions.

Spotting similar cases at a client

An ITSM implementation partner can usually tell in one conversation with the service delivery director whether a client has cases like this.

  1. Post-incident reviews cite change numbers. If reviews name the changes involved, collisions can be found later.
  2. Conflict results are stored, not just displayed. Ask whether conflict checks are saved on the change record.
  3. Change events reach the on-call tool. An ITSM-to-on-call integration means recent-change correlations exist on past incidents.
  4. Emergency rollbacks get their own approval. A separate emergency change or back-out task with an approver makes step 6 attributable.
  5. CMDB relationship history is kept. Without it, nobody can tell whether a dependency existed on the night.
  6. There is a real date ahead. An ITSM consolidation, an on-call tool switch or a managed services handover puts the history at risk.

Note metadata only: tools, years, rough counts of change-related major incidents, owners. Do not export incidents, take screenshots or pull samples. The records across systems map shows how these records join, and the who can license answer covers whom to introduce. For ticket-level context, see ServiceNow incident data.

Qualification and what stays open

The company needs 50+ full-time employees at peak (contractors excluded), a documented operating history, rights to license the records, and an authorized contact. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. Partners sign the Channel Partner Agreement before submitting introductions.

What stays unknown until SourceX reviews it: buyer interest in collision history, how client contracts affect records about services run for customers, which years are selected, and whether an offer follows.

Next step

Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

If a client's reviews can still trace a collision from the conflict check to the CMDB fix, register as a partner, check the company on who qualifies, and read the program terms before you introduce the service delivery director.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Is one collision enough to make a company interesting?

One case shows the pattern, but the value is in a history of many: different collisions, different rollback choices and the process changes that followed. Ask roughly how many major incidents were linked to changes over the years the records cover.

What if the post-incident review lives in a wiki rather than the ITSM tool?

That is common. Note where reviews are kept and whether they cite the incident and change numbers. A review that names both can still be joined to the ITSM records later.

Sources

  1. docs.servicenow.com/bundle/paris-it-service-management/page/prod…
  2. support.pagerduty.com/docs/change-events
  3. support.pagerduty.com/main/docs/servicenow-change-requests-integra…
  4. refer.sourcex.si/terms

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment