About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
A backup status is the backup software's account of what happened during one run. It is not a verdict on whether you can recover. "Success" means the software believes it finished what it was asked to do. "Warning" and "partial" mean it finished some of it, or finished it with a caveat. "Failed" means it stopped. 'Missed' is different from all four: it is a run that never started, so there is usually no status to read.
For a customer receiving weekly or daily reports, the practical rule (our recommendation) is this. Treat failed and missed as needing action today. Treat warning and partial as needing a prompt first assessment, ideally the same working day the report arrives, because the vendor definitions below show they can hide a real gap: skipped files, data that may not be consistent, or a transfer that did not finish. The assessment checks which items were affected, when the last usable restore point was made, and whether that still meets your agreed recovery point target (how much recent work you can afford to lose). Only the repair deadline is then set by risk, and a gap in an important system is not left for days. Treat success as a good sign that still has to be proven by a restore test, which is covered in a separate article on why a successful backup job is not a successful restore.
Why it matters
Reports are the only routine evidence most businesses see about their backups. If nobody reads them closely, a repeated warning can run for months before anyone learns that one folder, one database or one disk has not been protected.
The labels are also not standard. The same word can mean different things in different products, and some products use words the others do not have. This guide compares how three vendors describe their own statuses. It does not cover every backup product, and your provider's report may use other labels. The comparison shows the questions to ask rather than a fixed dictionary.
Success
Veeam's documentation defines Success for a task as "the task is completed successfully" (Veeam, Viewing Job Session Results). Note that Veeam counts results per task, "one task per one VM", so a job covering ten machines can contain ten separate results.
AWS Backup treats a completed job as the healthy outcome, and its dashboard counts completed jobs and completed-with-issues jobs together as its "successful" rate line (AWS Backup console dashboards). That detail matters: in that chart, a job with a warning-like message is still counted on the successful side.
What success does not tell you (our reading): whether the data inside the backup is usable, whether the right items were selected in the first place, or whether a copy exists somewhere other than the system that was backed up.
Warning, or "completed with issues"
This is the most ambiguous label, because vendors use it for several unrelated situations.
- Veeam: a warning means "the task is completed with minor errors". The vendor adds: "Depending on the nature of the errors, the backup data may not be consistent." (Veeam). So the same orange mark can mean a trivial message or data that cannot be relied on.
- AWS Backup: "Completed with issues is a status of a job displayed in the console that denotes a completed job with a status message." The AWS documentation also says this status exists only in the console: "it cannot be tracked via CloudWatch" (AWS Backup metrics with Amazon CloudWatch). If your reporting is built from alerts rather than the console, these jobs may not appear (our reading of that statement).
- Azure Backup, file and folder agent: a job completed with warnings "means that at least one file wasn't able to be backed up". Microsoft says the job skipped that file and backed up the rest, and lists causes such as unsupported file attributes, a file system issue, or another process such as antivirus software holding files open (Microsoft, Troubleshoot the Azure Backup agent).
- Azure Backup, virtual machines: Microsoft's job-status table shows a "Completed with warning" status in a case where the snapshot step finished but the transfer to the vault did not (our reading of the table's columns) (Microsoft, Back up Azure VMs in a Recovery Services vault). A snapshot is a point-in-time copy held near the original system. If it is not in the vault, then the protection may still depend on the same platform that holds the original.
The lesson is that a warning must be opened promptly, not counted. The question for each warning is: what exactly was not protected, and is that item important? A warning is not safe to leave for later until that question has been answered.
Partial
"Partial" is rarer and less consistently defined. AWS lists PARTIAL among the valid states of a backup job in its API documentation (AWS Backup, ListBackupJobs), but the pages we read define "partial" only for recovery points: the count of recovery points that AWS Backup "started to create but could not finish". AWS says it retries later and "retains the partial recovery point" (CloudWatch page, above). AWS's dashboard also leaves partial jobs out of its health percentages, which it says means totals "may not equal 100%" (dashboards page, above).
Our reading: a partial restore point is a copy that exists but should not be assumed complete. In a report that mixes "partial" with "success", ask whether partial points are counted in the "protected" figure.
Other products use "partial" loosely for a run where some items succeeded and others did not. Whatever the product calls it, ask for the list of items that were not captured.
Failed
Veeam defines it as "the task is not completed due to a blocking error". AWS lists Failed and Expired as the two "unsuccessful" lines on its dashboard, and groups Failed, Expired and Completed with issues together as "problematic" jobs (dashboards page, above).
A failed job summary does not on its own tell you what happened to each item. Veeam, for example, reports per task ("one task per one VM"), so a job covering several machines can contain tasks that succeeded alongside tasks that failed. AWS Backup likewise lists a PARTIAL job state and keeps partial recovery points. Do not assume that a failed run produced no new restore point, and do not assume it produced a usable one. Check the item-level results: which machines, folders or databases got a new restore point, which did not, and which hold a leftover incomplete one (our reading; products differ). What matters for recovery is then how long it has been since the last good restore point for each affected item, and whether the failure is a one-off or has repeated.
Azure's documentation gives a reason to check the age of the last good point rather than trust a single line. On a vault's backup-item list, Microsoft says you get "the last backup status for each item, latest restore point available" (Microsoft, Monitor Azure Backup protected workloads). The latest restore point is the figure that measures what you could recover to today.
Missed
'Missed' is our label for a run that was scheduled and did not start. It deserves its own heading because it fails differently: nothing runs, so nothing is reported, and a report built only from job results will look clean.
Vendors have related states, though not under that name:
- AWS Backup has an Expired status. "A backup job changes from status CREATED to EXPIRED if a backup cannot begin within the start window time." That is a scheduled job that was not able to start in time (CloudWatch page, above).
- Azure's file and folder agent documentation has a section for scheduled backups that do not trigger automatically even though manual backups work, with causes such as a disabled scheduler task (Azure agent troubleshooting page, above).
There is also the mirror problem: a report that shows activity that is not real. Microsoft documents that if a file and folder agent job crashes, "the status might not get propagated to Azure", so the portal can show the job as in progress, and the stale entry is removed automatically after 30 days (Azure agent troubleshooting page, above). A job shown "in progress" for days is therefore a question to ask, not a sign of a very large backup.
To detect a missed run, check the date of the latest restore point for every protected item, not just the list of jobs that ran.
A triage rule you can apply to a weekly report
This is our recommendation, not vendor guidance.
| Status | What to do | Within |
|---|---|---|
| Success | Spot-check that the list of protected items matches what you expect; keep restore tests scheduled | Routine |
| Warning / completed with issues | First assessment: obtain the warning message, the affected items and the last usable restore point, and compare it with your recovery point target. Then set a repair deadline by risk and get a fix or a written acceptance | Assess promptly (same working day); repair deadline set by risk, today if the affected item has no recent usable point |
| Partial | First assessment: obtain the list of items or restore points that are incomplete; confirm a recent complete point exists for each and that it meets your recovery point target. Then set a repair deadline by risk | Assess promptly (same working day); act today if it leaves no recent usable point |
| Failed | Ask for the item-level results, the cause, the date of the last good restore point for each affected item, and when the next run is expected to succeed | Today |
| Missed (no run, expired, or stale in-progress) | Ask why it did not start; confirm the age of the latest restore point | Today |
Two cross-checks improve this table. First, repeated warnings on the same item are a failure in slow motion, so look at several weeks together as well as assessing each warning when it arrives. Second, compare the report's list of protected items against your own inventory of systems; a system that is not listed at all can never produce a warning.
Questions to put to your provider
- For each status label in our reports, what is the exact definition, and which can still leave a usable restore point?
- Do warnings and partial results appear in the report with their message text, or only as counts?
- Is a skipped or excluded file or database listed by name?
- How is a missed or expired run detected, and who is told?
- Does the report show the age of the latest good restore point for each protected system?
- Who reviews the reports on your side, how often, and what is the written follow-up when a status is not success?
- Are reports produced from the job results alone, or are they compared against the list of systems that should be protected?
Read-only checks
These are checks you can do from a report or a console without changing anything. No commands are included in this article, and nothing here was tested.
- Open the most recent four weeks of reports and mark every item that is not plain success.
- For each marked item, find the message text and the restore-point date.
- Confirm that each critical system has a restore point newer than your agreed interval. Whether that interval is 24 hours or a week is your decision; the report cannot supply it.
- Check that a system you added recently appears in the report.
A worked example (illustrative)
The following is invented. A small firm receives a weekly report with 38 green ticks and two orange ones. Nobody opens the orange ones because the summary line says '38 of 40 successful'. A month later a staff member deletes a project folder. The firm asks for last Tuesday's version and learns that the file server job has completed with warnings every night for five weeks, because a large folder was held open by another program and was skipped each time.
Nothing was lost beyond that folder, but the recovery took a day of reconstruction. With the triage rule above, the first orange tick would have been opened the same working day, the message read, the skipped folder and the last usable point for it identified, and the exclusion fixed within a deadline set by how important that folder was, or the risk accepted in writing.
Read next
Related articles you may want next.
- Backup & RecoveryGrandfather-father-son retention: counting the restore points a policy really keepsHow to count the restore points a 30 daily, 12 monthly backup policy keeps, where the gaps fall, and what to ask a provider to confirm in writing.Read article
- Backup & RecoveryA successful backup job is not a successful restoreWhy a finished backup job does not prove you can restore, and which checks show the copy can actually be opened and used.Read article
- Backup & RecoveryBackup, replication, failover and disaster recovery: four terms a buyer should not confuseBackup, replication, failover and disaster recovery cover different failures. See which protects against deletion, ransomware, outages and site loss.Read article
Risks and limits
- This guide covers three vendors' documentation. Your product may use different labels, and vendors change them between versions.
- We did not test any product, and we could not find a single vendor page that defines all five of these words together.
- 'Missed' is our term. We have not seen a vendor define it under that name in the pages we read.
- A clean report is not evidence of recoverability. Only a restore test provides that.
Next step
Take last week's report and classify every status using the triage table, then ask your provider in writing for the message behind each warning and the restore-point age of each critical system. If you are reviewing a hosted backup service, the Backup and Recovery page is a starting point for that conversation; ask for the scope in writing.
Sources
- AWS Backup console dashboards, Amazon Web Services; date unavailable.
- AWS Backup metrics with Amazon CloudWatch, Amazon Web Services; date unavailable.
- ListBackupJobs, Amazon Web Services; date unavailable.
- Viewing Job Session Results, Veeam Software; page updated 2026-07-21, applies to build 13.1.2.57.
- Troubleshoot the Azure Backup agent, Microsoft; last updated 2025-12-10.
- Back Up Azure VMs in a Recovery Services Vault, Microsoft; last updated 2026-03-25.
- Monitor Azure Backup protected workloads, Microsoft; last updated 2026-01-30.
All accessed 15 November 2026.
Learning notes
- A backup status describes the run, not your ability to recover.
- A warning is a question; open it promptly, read the message and check the affected items and last usable restore point.
- A failed job summary does not say what happened to every item; check item-level results.
- "Completed with issues" may be counted on the success side of a chart, so read the detail.
- A missed run leaves no failure to report; check the date of the latest restore point instead.
- The age of the newest good restore point is the number to watch.
This article was originally published on .