Separate true requirements from implementation issues before migrating Jira. Jira's issue keys and links are useful identifiers, but “done” tasks, defects and epics do not all become requirements. Decide which system will own each record after migration; Jira may remain the delivery authority.
Which Jira records are requirements?
Inventory projects, issue types, custom fields, parent relationships, link types, versions, comments, attachments and installed requirements apps. Mark each issue as an approved engineering requirement, proposed requirement, implementation task, defect or other record. Preserve Jira keys as source references even if the destination assigns new IDs.
Which Jira query should we export?
Atlassian documents Jira Cloud CSV export, including an all-fields option and comments. It also warns that bulk physical attachments are not natively downloaded with that CSV. Confirm the exact Jira deployment and export permissions. Freeze or timestamp the query so a later issue update does not make counts incomparable.
What does each mapped record mean?
Translate requirement text, rationale, owner, state and relationships into the destination's engineering model. For a synthetic flight-software requirement, keep a Jira implementation task linked to the approved requirement instead of duplicating the requirement as another task. Reconcile issue links by type and direction; record app-specific semantics before the app is removed.
How do we preserve decisions and evidence?
Comments may contain decisions, but an exported comment thread is not automatically a formal review. Decide which historical discussions are archived and which current approvals must be re-established. Test links to test results and attachments separately, then have engineering owners approve the destination baseline.
What should the Arc import preserve?
Jira is on Arc's approved direct import list. That does not remove the need to classify source issues or review the generated graph. Keep Jira for delivery work if it is the right authority; define how an engineering change updates or relates to its tasks.
Frequently asked questions
Should every Jira issue become a requirement?
No. Classify requirements, tasks, defects and decisions before mapping them to destination record types.
Does Jira Cloud CSV include attachment files?
Atlassian says the CSV can carry attachment URLs, but Jira Cloud does not natively bulk-download the physical files through that route.
Can Jira remain after Arc migration?
Yes. Jira can own implementation work while Arc owns approved requirements and verification, with a defined link and reconciliation rule.
Should Jira issue keys be kept if the destination creates new IDs?
Keep each Jira key as a source reference and reconcile it to the destination record. Classify the source issue and preserve the type and direction of its links, including installed-app meaning, so an implementation task remains related to the approved engineering requirement rather than becoming a duplicate authority.
How should exported Jira comments be used for acceptance decisions?
Archive relevant historical discussions with their source context. An exported comment thread does not automatically become a formal review or approval; identify which current decisions must be re-established and have engineering owners accept the mapped baseline and evidence links.