Veeam Agent — Corrupt Install: Force Removal & Clean Redeploy (Error 1605)
Audience: T2 / T3 Version: 1.0 Last Updated: September 2026 Applies To: Veeam Agent for Microsoft Windows (v12 agents on current Windows 11 builds), Veeam Backup & Replication v12 → v13
Overview
This guide covers a corrupt Veeam Agent for Microsoft Windows install that can be neither repaired nor uninstalled, and the follow-on failure where the B&R server keeps a stale record of the removed agent and every rescan fails with Windows Installer error 1605.
Common symptoms:
- Veeam Agent (v12) on the endpoint will not install, repair, or uninstall — installer rolls back or errors out
- B&R rescan of the managed computer warns:
Failed to connect to <host> — This action is only valid for products that are currently installed. Failed to invoke rpc command. 1605/ERROR_UNKNOWN_PRODUCTin Windows Installer logs on the endpoint- Seen after the v12 agent era on current Windows 11 builds; upgrading the B&R server to v13 does not clear it by itself
Root cause (two layers):
- Endpoint: the agent's MSI registration is corrupt, so Windows Installer can neither service nor remove the products. The only way out is manual de-registration (below).
- B&R server: after the endpoint is force-cleaned, the B&R configuration database still holds the old agent/managed-computer record. A passive rescan discovers but does not deploy — it tries to manage the package its database says is installed, the endpoint's Windows Installer answers 1605 (unknown product), and the rescan surfaces it as a connect/RPC failure. Another rescan will never fix this; the stale record has to go.
Part 1 — Force-remove the corrupt agent (endpoint)
⚠️ Back up the registry keys before deleting them. Export each key you touch to a dated folder (e.g. C:\Users\<user>\Desktop\veeam-reg-backup).
1.1 Delete services — order matters
Delete VeeamDeploySvc (Veeam Installer Service) FIRST. It self-heals and re-registers the other components; removing it last means it can put back what you just deleted.
# 1. Installer service first (self-heals if left for last)
sc.exe stop VeeamDeploySvc
sc.exe delete VeeamDeploySvc
# 2. Then the agent and transport services
Get-Service Veeam* | Select-Object Name, Status # enumerate what's present
sc.exe stop VeeamEndpointBackupSvc
sc.exe delete VeeamEndpointBackupSvc
sc.exe stop VeeamTransportSvc
sc.exe delete VeeamTransportSvc
(Service names vary slightly by version — delete whatever Get-Service Veeam* returns, installer service first.)
1.2 De-register from Windows Installer — all three registration points
The corrupt install is registered in three places; all three must go or msiexec and the B&R deployer keep finding the ghost product.
Product GUIDs for the v12 agent generation this was written from (verify against the endpoint's own registry before deleting — GUIDs are version-specific):
| Component | Uninstall-key GUID | Packed code (Installer\Products) |
|---|---|---|
| Veeam Agent for Windows | {7796202E-3320-41ED-9A2C-14613AEED3D3} |
E20269770233DE14A9C24116A3EE3D3D |
| Veeam Backup Transport | {A2D1052D-657D-48E3-8D7C-A77C8B07A665} |
D2501D2AD7563E84D8C77AC7B8706A56 |
| VeeamDeployerService | (registered as its own uninstall entry) | — |
- Uninstall keys —
HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\{GUID}(and theWOW6432Nodeequivalent) for the Agent, Transport, and VeeamDeployerService entries. - Installer product registrations —
HKLM\SOFTWARE\Classes\Installer\Products\<packed code>for both packed codes above. - MSI UserData —
HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Installer\UserData\S-1-5-18\Products\<packed code>\InstallProperties(delete the product node).
1.3 Scheduled tasks and reboot
- Remove Veeam scheduled tasks:
Get-ScheduledTask | Where-Object TaskName -like '*Veeam*' | Unregister-ScheduledTask -Confirm:$false - Reboot the endpoint. A half-present service that hasn't been fully released produces exactly the partial-connect + 1605 pattern on the next rescan.
1.4 Verify the endpoint is clean
Get-Service Veeam* # nothing
Get-ScheduledTask | Where-Object TaskName -like '*Veeam*' # nothing
Get-ChildItem 'HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall' |
Get-ItemProperty | Where-Object DisplayName -like '*Veeam*' # nothing
All three must return nothing before touching the B&R side.
Part 2 — Clear the stale record on the B&R server
Do not just rescan again. The rescan is what's failing; it queries the product the database still believes is installed.
- In the B&R console: Inventory → Physical Infrastructure → the protection group containing the endpoint.
- Right-click the computer → Remove from configuration (removes the stale agent/managed-computer record; existing backups stay in the repository and can be re-mapped).
- If the computer was added as an individually named member, re-add it; if the protection group discovers by AD/CSV, run Rescan on the group so it's rediscovered as a new computer.
- Right-click the rediscovered computer → Install agent (an explicit deploy — discovery alone does not necessarily push the agent).
If deployment still fails: rule out transport/RPC
Failed to connect … Failed to invoke rpc command can also be plain connectivity. From the B&R server against the endpoint:
Test-NetConnection <endpoint> -Port 135 # RPC endpoint mapper
Test-NetConnection <endpoint> -Port 445 # SMB (admin$ / C$)
Test-Path \\<endpoint>\admin$ # admin share with the B&R credential
Get-WmiObject Win32_OperatingSystem -ComputerName <endpoint> # WMI/DCOM
Test-NetConnection <endpoint> -Port 6160 # Veeam Installer Service (once deployed)
Also confirm on the endpoint: Windows Firewall allows File and Printer Sharing, WMI, and Remote Service Management; and the credential B&R uses for it is a local administrator.
Verification
- Protection group rescan completes green — no 1605, no RPC warnings.
- Endpoint shows the current (v13-generation) agent version under the protection group.
- On the endpoint,
Get-Service Veeam*shows the new agent services running. - Run the backup job (or wait for its schedule) and confirm the first backup session succeeds — deployment green is not the finish line; a completed restore point is.
Quick Reference
Endpoint: delete VeeamDeploySvc first → delete agent/transport services → de-register all three registry points (Uninstall keys, Classes\Installer\Products, MSI UserData InstallProperties) → remove scheduled tasks → reboot → verify clean.
B&R server: remove the computer from configuration → re-add/rediscover → explicit Install agent → verify green rescan + first successful backup.
Rule of thumb: 1605 on rescan after a manual agent removal = stale B&R database record. Remove-and-re-add, don't re-rescan.
Related Documents
- Adding & Replacing Computers in Veeam BDR
- Agent & Endpoint Offline
- Veeam v13 Fleet Upgrade — Runbook & Milestone Tracker — agent currency is milestone M5; this failure mode blocks M5
- Veeam & BDR Troubleshooting Guide
Document History
| Version | Date | Author | Changes |
|---|---|---|---|
| 1.0 | September 2026 | Apis | Initial creation from a live incident (corrupt v12 agent force-removed; v13 B&R rescan then failed 1605 against the stale managed-computer record). Related to the v13 fleet upgrade (HALO 1179664). |
Confidential — Internal Use Only