OPERATIONS AND RECOVERY / FACTORYTALK ASSETCENTRE
Your controller backups have a server. Who can reach it?
The machine you depend on after an incident deserves its own access and recovery check.
Ask your controls team where the known-good copy of a controller program lives. Then ask who can change that copy.
For organizations using Rockwell Automation FactoryTalk AssetCentre, those questions lead to an industrial asset-management server, its database, and the engineering machines connected to it. A backup application is part of the recovery plan. It is also a system with accounts, services, and software that need protection.
This review starts with the recovery dependency. You do not need Deception Check telemetry or a new security product to use it.
NORMAL RELATIONSHIPS / WHAT THE RECOVERY SYSTEM CONNECTS
- AssetCentre server
Stores project versions and coordinates backup work.
- Engineering agents
Workstations exchange project data and device tasks.
- Controllers
Run the programs that operate equipment.
Bring back four answers, with evidence
Start with a read-only inventory and a conversation with the engineering owner. Review configuration and existing records; do not test an exploit against an operating plant.
Find the system and its owner
List the AssetCentre server, database, agents, and engineering owner. Include restored VMs and older standby copies.
KEEP: inventory, owner, installed version
Verify the fix and the path to it
Check the running installation against PN1559. If an upgrade must wait, name the owner of the temporary access restrictions and a review date.
KEEP: version evidence, change record, exception
List who can reach the services
Check firewall rules, remote support, client access, and trusted engineering hosts. An internal address is not an access policy.
KEEP: permitted clients and reviewed rules
Prove you can recover separately
Identify a protected project copy and the credential custodian and access procedure needed to recover it. Test restoration in an approved isolated environment.
KEEP: restore result, copy location, access owner
TAKE THIS TO THE ENGINEERING OWNER
A two-page worksheet for the review
Record the version, allowed clients, recovery copy, evidence locations, and owners. The second page separates suspicious requests from confirmed changes and sets a safe recovery exercise.
Download the worksheet PDFOpen the readable worksheetNew attention to a known fix
A newly published research repository claims a proof of concept for CVE-2021-27460 and CVE-2021-27466. A proof of concept is code intended to demonstrate a weakness. We inspected its published source, but did not run or independently validate it. That is a reason to revisit exposure, rather than evidence of a new attack campaign.
- APRIL 1, 2021The vulnerabilities and fix
Rockwell published PN1559 covering nine AssetCentre vulnerabilities.
- OCTOBER 31 EASTERN / NOVEMBER 1 UTC, 2026A research repository appears
The project names two deserialization flaws from that advisory.
Rockwell's advisory identifies v10.00 and earlier as affected and v11 or above as the remediation. Treat v11 as the historical fix floor, then confirm the supported target release and upgrade path for your installation. The advisory also points to IPsec guidance, QA46277, for restricting unauthorized clients when an immediate upgrade is not possible. Restrictions reduce exposure; they do not replace the fix.
The two CVEs were absent from the official CISA Known Exploited Vulnerabilities catalog checked on November 4. We found no confirmed reporting of their use in the wild in the sources reviewed. The practical decision is still based on your version and reachable services.
What the technical path actually means
Claroty Team82's original research describes AssetCentre's central server, database, and agents running on engineering workstations. Those agents communicate with automation devices. That relationship makes the recovery application worth reviewing alongside the controllers.
The two named flaws concern unsafe deserialization: software rebuilding an object from received data without sufficiently checking it. In the vulnerable implementation, supplied data may let a remote caller execute commands without logging in. CVE-2021-27466 concerns ArchiveService.rem; CVE-2021-27460 describes the wider remoting issue. Remoting is how software components communicate across processes or machines.
The wider advisory contains other flaws, including SQL injection and command injection. They are separate issues; the new repository does not demonstrate every route or a complete path to a controller.
POTENTIAL ATTACK PATH / REVIEW EACH BOUNDARY
- 01 / REACHContact a vulnerable service
A network path exists from the caller to the server.
- 02 / EXECUTEAbuse object handling
Unsafe deserialization can turn supplied data into code execution.
- 03 / CHANGEAffect recovery assets
Review the server, stored projects, and service-account reach.
- 04 / PROGRESSAttempt device access
Further access depends on agents, privileges, and controller protections.
For a nontechnical owner, the consequence is straightforward: someone who gains control of the backup application may be able to affect the material you plan to restore. Ask engineering to establish the actual permissions and device reach before assigning a plant-wide impact.
Hunt across the boundaries, not just the filename
Build the case from existing access, execution, change, and device records. Preserve originals and their time zones, then correlate on host, account, destination, and time. A source address alone is not a person or an authorization decision.
| Question | Ordinary source | Keep with the case | What it can establish |
|---|---|---|---|
| Service request | IIS access logs where the remoting endpoint uses HTTP; firewall or network records for other configured transports. | Caller, destination, time, method, path, status, and allowed-client context. | A request reached a recorded service. It does not establish execution. |
| Process execution | Existing endpoint detection and response (EDR) process telemetry, Sysmon Event ID 1, or Security Event 4688 with the relevant auditing enabled. | Host, parent and child process, account, command line when collected, and nearby activity. | An execution record can corroborate follow-on behavior. Normal maintenance can look similar. |
| Project or archive change | AssetCentre audit/version records and approved change tickets, plus file or database auditing if configured. | Changed object, identity, before/after version, approval, and independently retained copy. | Establish what changed and whether it was authorized. A process event alone cannot answer that. |
| Controller operation | Engineering and controller records, compared with the operations team's work log. | Device, operation, account, project version, time, and the operator who confirmed it. | A separate device-side check is needed to establish a download or operational change. |
Availability depends on your logging and retention settings. Windows 4688 command lines require separate policy configuration; IIS logs cover HTTP endpoints, not every .NET remoting transport. These are collection and triage instructions, not a tested signature for the new PoC.
Three useful hunts for the team this week
- Unexpected callers: compare recorded remoting clients with the approved engineering-host list. Investigate a new source or a remote-support host reaching the server outside its approved work.
- Unexpected execution: on systems where an EDR agent is supported, review unusual child processes and scripts from the relevant server services. Correlate with maintenance before escalating.
- Unexpected project changes: compare archive versions, restore activity, and device downloads with approved work. Include deletion, replacement, and missing audit periods in the review.
Do not reduce this to blocking .rem files. That extension can identify legitimate remoting services, and its presence alone is not an exploit. Start by scoping the host and its approved clients. Keep event collection separate from any claim of confirmed compromise.
Recovery needs a second place to stand
A second folder on the same server may share the same failure and access risks. Our recommendation is to retain a protected recovery copy with separately controlled access, document the restore dependencies, and test it without touching production equipment.
Choose one recent controller project. Have the engineering owner recover it in an approved isolated test environment using the retained copy and the documented credentials. Record what worked, what was missing, and who owns the next action. A backup job's green status is useful; a successful restore answers a different question.
Could your team recover a controller project if the AssetCentre server and its usual accounts were unavailable? That is the question to take into the next operations review.
Sources and scope
This is a practical review prompted by newly surfaced source code for known 2021 flaws. It contains no Deception Check capture of these CVEs and makes no claim of FactoryTalk-specific decoy coverage.
- Rockwell PN1559: affected versions and remediation
- Claroty Team82: architecture and original research
- CISA ICSA-21-091-01: original advisory
- Research repository, pinned to the source inspected
- CISA Known Exploited Vulnerabilities catalog
- Microsoft Sysmon event documentation and Windows process-creation auditing
- Microsoft IIS logging configuration
Repository and exploitation status were checked on November 4. Historical dates, current reporting, and the recommended exercise are identified separately above. Engineering approval is required for changes to an operating control environment.
