Upgrades and patching in Administration

Upgrades and patching

Family releases, patches, skipped records and how to plan an upgrade that does not eat a month.

Quick answer

Family releases, patches, skipped records and how to plan an upgrade that does not eat a month.

Key takeaways

Upgrades are routine, plan them into the calendar
Skipped records are decisions, not noise
Automated tests pay for themselves at upgrade time
Rollback means restore, so test thoroughly before production

Release cadence

ServiceNow ships two family releases a year plus regular patches and hotfixes. Instances must stay within supported versions, so upgrades are routine work rather than projects.

Skipped records

During upgrade the platform keeps your customisation of any record you changed and marks the shipped version as skipped. Reviewing skipped records is the real work of an upgrade: for each one, decide whether to keep your version or take the new one.

A workable plan

Upgrade sub production first and give testers a fixed window.

Upgrade dev, review skipped records, then upgrade test
Run ATF suites before and after so regressions surface early
Prioritise skipped records on high traffic tables and on anything security related
Schedule production for a low volume window and keep the rollback conversation honest, rollback means restore from backup

Practice challenge

+0 XPStreak ×0

Question 1 of 2

Where do you review changes the upgrade did not apply?

Frequently asked questions

What is a skipped record?

A record where your instance has a customisation, so the upgrade left your version in place and parked the new one for review.

How long should an upgrade take?

The technical upgrade of an instance is measured in hours. The review and test cycle is what varies, and it shrinks when customisations are few and automated tests exist.