Skip to content

Backup succeeded but restore fails: VSS writer “Failed” state

  • Home
  • Blog
  • Backup succeeded but restore fails: VSS writer “Failed” state
Backup succeeded but restore fails: VSS writer “Failed” state
Short answer

If your backup software reports “successful” or “completed with warnings” but the restore fails or the database comes back inconsistent, the problem is most likely in the VSS writer layer. A backup taken while a writer is in the “Failed” state is not application-consistent; at best it is crash-consistent. The fix is to find the failed writer with vssadmin list writers, restart its hosting service in a maintenance window, check the shadow copy storage, and prove the backup with a real restore test.

Symptoms: “State: [9] Failed” and silent failure

  • The backup job shows green or “Completed with warnings”. The log contains VSS or writer warnings, but nobody reads it regularly.
  • In an elevated command prompt, vssadmin list writers shows “State: [9] Failed”, “State: [5] Waiting for completion”, or “Last error: Retryable error” or “Timed out” for one or more writers.
  • SqlServerWriter is missing from the list or in a failed state. The SQL Server VSS Writer service is stopped or disabled.
  • The restored SQL Server database will not open, gets stuck in recovery, or DBCC CHECKDB reports inconsistencies.
  • The restored server runs a disk check at boot, and some applications are missing their latest transactions.
  • Older shadow copies disappear on their own; vssadmin list shadows shows fewer copies than expected.

Root cause: when a writer fails, the backup stops being application-consistent

The Volume Shadow Copy Service (VSS) coordinates three components: the requester (the backup software), the writer (the module that makes the data of components such as SQL Server, Exchange, Active Directory and the registry consistent), and the provider (the layer that creates the shadow copy). Writers complete open transactions, roll transaction logs and flush caches to disk. VSS then briefly freezes application write I/O. According to Microsoft, the application freeze is not allowed to take longer than 60 seconds and the provider must commit the shadow copy within 10 seconds; if these limits are exceeded, shadow copy creation can be aborted.

The real danger is this: when a writer fails, many backup products do not stop the job. They continue without that writer and mark the job “completed with warnings”. The resulting backup is a copy of whatever blocks were on disk at that moment, in other words crash-consistent. Data still in memory and not yet written to disk is not in the copy. The restored database has to recover from its own transaction log as if it were starting after a power cut, and that does not always succeed. Microsoft’s SQL Writer documentation states it plainly: if the SQL Writer service is disabled, any utility that relies on VSS snapshots can break and might take backups of databases that are not consistent.

The second root cause is shadow copy storage (the diff area). The built-in Windows system provider uses copy-on-write: after a shadow copy is taken, changed blocks are first copied into this area. When the area runs out of space, shadow copies are deleted, beginning with the oldest. If the limit is too small, a shadow copy cannot survive heavy write activity and older copies quietly vanish.

Writers usually get stuck in a failed state because of an interrupted backup job, a hung service, or a fault in the application that hosts the writer. The known way to clear it is to restart the service that hosts the writer; if that is not enough, the server needs a planned restart.

Step-by-step fix

  1. Open an elevated command prompt and record the writer states. A healthy writer shows “State: [1] Stable” and “Last error: No error”.
  2. Note the name of the failed writer and identify the service it depends on. SqlServerWriter belongs to the SQL Server VSS Writer service (SQLWriter). The System Writer runs inside the Cryptographic Services (CryptSvc) service. For other writers, check the service of the application that provides the writer.
  3. Review VSS errors in the Application event log; they usually show at which phase (freeze, thaw, backup complete) the writer failed.
  4. After confirming that no backup job is running, restart the relevant service and list the writers again.
  5. Check the shadow copy storage and existing copies. If the limit is too small for the daily change rate of the volume, increase it.
  6. For SQL Server, verify that the SQL Writer service runs as Local System and that the NT Service\SQLWriter login exists in SQL Server; the service needs sysadmin rights because it briefly freezes all I/O for the instance during a snapshot.
  7. Start a new backup manually and confirm that the job log no longer contains writer warnings.
  8. Finally, actually restore the backup: restore the database to a test server under a different name and run DBCC CHECKDB.
vssadmin list writers
vssadmin list shadowstorage
vssadmin list shadows

# PowerShell: service state and recent VSS events
Get-Service -Name SQLWriter, VSS, CryptSvc
Get-WinEvent -FilterHashtable @{LogName="Application"; ProviderName="VSS"} -MaxEvents 20

# Restart the writer service while no backup job is running
Restart-Service -Name SQLWriter
vssadmin list writers

Before restarting services, make sure no backup job is running; stopping a writer service in the middle of a job breaks that job. Microsoft’s warning also applies to the resize command below: resizing the shadow copy storage association may cause existing shadow copies to disappear.

vssadmin resize shadowstorage /For=C: /On=C: /MaxSize=20%

For restore testing, RESTORE VERIFYONLY alone is not enough. According to Microsoft, it checks that the backup set is complete and readable, some page header fields and the checksum if present, but it does not attempt to verify the structure of the data in the backup. The real proof is DBCC CHECKDB running against a restored copy.

RESTORE VERIFYONLY FROM DISK = 'D:\Backup\ERP.bak';
-- After restoring to the test server as ERP_Test:
DBCC CHECKDB (N'ERP_Test') WITH NO_INFOMSGS;

Permanent prevention

  • Treat “completed with warnings” as a failure. Monitor writer warnings separately in backup reports and check the output of vssadmin list writers regularly.
  • Do not disable the SQL Writer service and monitor its startup type. Because SQL Writer does not support transaction log backups, schedule SQL Server’s own full and log backups for databases in addition to VSS.
  • Set a shadow copy storage limit that matches the change rate of the volume and review it periodically with vssadmin list shadowstorage.
  • Apply the 3-2-1 rule: as defined by CISA/US-CERT, keep 3 copies of every important file (1 primary and 2 backups), on 2 different media types, with 1 copy stored offsite.
  • Put restore tests on the calendar: restore critical servers and databases to a test environment at fixed intervals and record the duration and the result.

When to call an expert

If writers return to a failed state even after the service and the server have been restarted, if a critical database comes back inconsistent from a restore, or if you do not have a single tested backup, trial and error only increases the risk of data loss. For the VSS and backup problems we encounter in the field at businesses in Denizli, Workas provides writer diagnostics, backup architecture reviews and the setup of regular restore tests as part of its server and backup solutions service. Our team works Monday to Saturday, 08:00–18:00, and provides priority response in emergencies.

Frequently Asked Questions

Can I use a backup marked “Completed with warnings”?

It may work at file level, but if the warning comes from a writer failure, the consistency of the application data is not guaranteed. Only trust that backup after verifying it with a real restore test.

Is a crash-consistent backup useless?

No. It contains the data on disk, and the application tries to recover from its own log at startup. However, in-memory and incomplete writes are lost, so it should not be the only protection for database servers.

Can I disable the VSS service to get rid of the error?

No. Microsoft advises against it; disabling VSS adversely affects all software that depends on it, such as System Restore and Windows Server Backup.

Can I take SQL Server log backups through VSS?

No. SQL Writer supports full, differential and copy-only backups, but not transaction log backups. Use SQL Server’s native BACKUP LOG for log backups.

Sources

  1. Volume Shadow Copy Service (VSS) – Microsoft Learn
  2. vssadmin – Microsoft Learn
  3. SQL Writer Service – Microsoft Learn
  4. RESTORE VERIFYONLY (Transact-SQL) – Microsoft Learn
  5. Data Backup Options – CISA / US-CERT

Let's Get Started

Let's Work Together!

Let's plan your company's IT needs together — we'll quickly find the right solution for you.

x

Since 2002, Workas Software Technology has provided businesses in Denizli with ERP, cybersecurity, network, server and backup, custom software and AI solutions.

Address
Kınıklı, Hüseyin Yılmaz Cd. No:67,
B Blok Z06, 20160 Pamukkale/Denizli, Türkiye
Get in Touch
0549 669 99 06
0535 089 29 29
Working Hours
Mon - Sat: 8:00 - 18:00