1. Define the original state and the rollback authority

Identify the service, the two endpoints, the approved maintenance window and the person who can authorize rollback. Name an executor, an independent checker where the site requires one, and the service owner who accepts the result. Include contact details in the controlled work record so the team does not need to find an approver during an outage.

Define what “original” means: the last verified connection map and service state immediately before this change. If the link is already faulty, document that baseline and agree the intended restoration target. Returning to the old ports cannot by itself promise healthy service. List any linked configuration, optic or equipment changes, with their own restoration owner.

2. Record both endpoints and every affected patch segment

Use an immutable connection ID for each affected segment. Record room, rack, panel, adapter position and equipment port at both ends, along with the patch-cord ID, fiber type, connector and polish at each end, length and relevant polarity mapping. For duplex or multifiber assemblies, document the actual fiber or lane mapping needed by the application rather than only the outer connector name.

Take overview and close-up photographs with references to the same connection ID. Photographs help locate equipment, but cannot replace a readable port map when labels are hidden. If a label conflicts with the drawing or live record, resolve the discrepancy before releasing that connection for change. Do not disconnect an unverified neighboring service to identify it.

Data center passive cabling

3. Keep the proposed change separate from the execution log

The original connection record is the restoration reference and should not be overwritten with the proposed arrangement. Use a separate change row to state the original endpoints, intended endpoints, action sequence and the expected result. Add an actual-action field for what was done, by whom and at what time.

Track intermediate states. A patch cord may have been removed from one end, replaced entirely, or moved only partly when the team stops. Rollback must start from the observed current state. Review the approved reverse sequence and its dependencies instead of assuming that every planned step has already happened. Preserve the original record even after the new connection works.

Open the fillable record template (save and print)

4. Prepare and reserve the restoration materials

Keep removed, serviceable patch cords identified and protected until the change is accepted and the agreed retention period ends. Record the storage location, connection ID, condition and the person responsible for release. A cord that has been damaged or contaminated during work needs assessment; “the original cord” does not automatically mean “ready to reconnect.”

If spares are used for rollback, match the approved fiber type, connector family, polish, polarity or lane mapping and usable route length. Include compatible inspection and cleaning tools, protective caps, labels and any approved adapters. Reserve the materials for the identified work rather than counting stock that another job may consume.

Fiber patching components

5. Set the rollback trigger and the last decision time

Agree measurable triggers before the maintenance window: an unresolved endpoint mismatch, failed agreed service checks, an unexpected impact on another link, or insufficient time to complete the remaining work safely. Record who evaluates each trigger and who authorizes the response. The team should not improvise a new acceptance threshold during the change.

Work backward from the permitted end of the window. Reserve time for restoring connections, verifying service and handling the agreed contingency. The latest rollback decision time is the window end minus those reserved durations. Estimate durations from the actual scope or a rehearsal; no single fixed allowance is suitable for every patching job.

6. Restore the recorded connections and verify the result

When rollback is authorized, stop further changes and reconcile the execution log with the actual connection state. Restore the approved original endpoints, fiber or lane mapping and routing using identified materials. Follow site optical-safety procedures; never look into a fiber or connector. Inspect mating end faces with suitable equipment, clean when needed and re-inspect before connection, including connectors that had protective caps.

Verify physical restoration and service restoration as separate results. Check the port map, labels and route condition, then perform the approved link and service checks. Where optical readings are required, record the measured value, instrument or device source, relevant settings and acceptance limit. A link indicator alone does not demonstrate the required application service.

If the original service target is not met, document the remaining discrepancy and escalate under the agreed plan. Do not close the work simply because every cord appears to be back in place. Keep the incident or change record open until the responsible service owner accepts the result or explicitly approves a revised disposition.

7. Close the record without losing the original map

Attach the final connection map, original reference revision, completed change log, restoration evidence, outstanding issues and acceptance names with timestamps. Record which spare materials were consumed and which original components remain reserved. This makes the next maintenance decision traceable.

Use the companion record set for the baseline, proposed and actual changes, rollback materials, triggers and restoration confirmation. Complete the plan before the work and update the execution fields during the window. The form is a coordination record; project-specific test limits and approval authority must come from the approved change plan.

Frequently asked questions