InnerWeb
Skip to main content

IT acceptance checklist after a system rollout: evidence for audit

IT acceptance after a rollout is the moment when the full set of audit evidence is created: the change trail, backups, email and error reporting. The list below shows what to collect, from which source and what to watch for when interpreting it, including the time zone of older entries.

107
entities covered by the audit trail
Source: card
95%
share of one machine table in the size of audit tables
Source: card

Why acceptance matters and what counts as evidence

IT acceptance closes the rollout from the environment side and gives the auditor a consistent set of documents. Evidence is not a declaration that something works, but a dated record that can be linked to a specific environment: an extract from a log, a test result, a release number, a confirmation of a backup.

The system delivers some elements of such evidence: the change trail, error reporting and an exchange log with the portal. They provide evidence for the audit, but do not replace the organisation's decisions: the plant assesses whether the scope of the record meets its requirements.

Change trail: scope and limits

The audit trail covers entities changed by a person, and entities written by the machine are excluded, because their history is source data. In the review the scope grew from 60 to 107 entities. One machine data audit table grew to 6331 MB, which is 95% of the size of the 59 audit tables together, and it documented no human activity, so it was excluded from the audit.

The time zone of the audit timestamps is Europe/Warsaw from October 2026. Before that a fixed offset shifted winter entries by one hour, and older entries were not recalculated. Record this information in the acceptance protocol so that the auditor does not treat the one-hour difference as a data error.

Share of the machine data table in the size of audit tables

Share of the machine data table in the size of audit tablesDonut chart: one audit table of machine data is 95% of the size of the audit tables, the remaining 58 tables are 5%One machine data table: 95 % (95%)Remaining 58 audit tables: 5 % (5%)100%One machine data table95 % (95%)Remaining 58 audit tables5 % (5%)
Source: Cell card: audit trail documentation, 6331 MB, 59 tables

Email and error reporting

The evidence for email is the delivery test result for system messages and a record that the SPF, DKIM and DMARC records remain unchanged. The plant's email is sometimes a source of false suspicions of spoofing when the sender comes from an external domain, so the description of the sender and the Reply-To is attached to the protocol.

For error reporting the evidence is a test event visible in the Sentry project, with a release version that is the short hash of the deployed image's commit and matches the tag in the registry. Reporting has frequency limits, so a single missing event does not always mean there was no error, which is why the test is repeated on a fresh event. Security alarms are not silenced.

Backups and exchange log

The evidence for backups is a record of the run, the storage location outside the instance host and the result of a trial restore. The exchange log with the portal records the direction, the event, the person, the result, the HTTP code and the time, and writing to the log never interrupts the main operation. An extract from this log is a natural piece of evidence for the integration.

How to put together the acceptance protocol

The protocol has one overview page and attachments: an extract from the audit trail, the email test result, a screenshot of the Sentry event, the backup confirmation and the exchange log extract. Each attachment has a date, an author and a release version. Also record the open points with an owner and a deadline, because an auditor values candour more than apparent completeness.

Checklist

Video

Frequently asked questions

Does the audit trail cover everything the system does?

No. It covers entities changed by a person. Entities written by the machine are excluded, because their history is source data.

Why do older entries have a one-hour difference?

The Europe/Warsaw zone applies from October 2026. Before that a fixed offset shifted winter entries by one hour, and older entries were not recalculated.

Does a missing event in Sentry mean there was no error?

Not always. Reporting has frequency limits, so the test is repeated on a fresh event. Security alarms are not silenced.

Sources

Let us prepare the IT acceptance protocol

We will help you put together the full set of audit evidence and describe the limits of the change trail, so that the acceptance is unambiguous.

Let's talk