FrostyGoop showed that an attacker does not need exotic exploits against a controller to cut heating to hundreds of buildings: a reachable Modbus TCP port and a flat network were enough. This profile covers the malware Dragos disclosed in July 2024, which was used in January 2024 against a municipal district energy company in Ukraine, leaving residents of over 600 apartment buildings without heat for almost two days in sub-zero temperatures. It draws on the Dragos intelligence brief, MITRE ATT&CK v20 (software S1165, campaign C0041), Nozomi Networks, CISA and press reporting of the Security Service of Ukraine's confirmation. It is written for OT security engineers, control system owners and SOC analysts who watch industrial networks, and it stays at behaviour level: no code, commands or configuration details.
Key takeaways
- It speaks a legitimate protocol. FrostyGoop reads and writes Modbus holding registers over TCP port 502, so its traffic looks like normal engineering traffic unless you know which hosts should be writing to which devices.
- The entry point was IT, not the controller. Dragos assesses the attackers likely exploited an externally facing router months before the attack, planted a web shell, and harvested credentials before reaching the heating controllers.
- Antivirus did not help at first. Dragos reports that no antivirus vendor flagged the binaries as malicious when they were found in April 2024, so detection has to rest on network behaviour.
- Loss of view lengthened the outage. The controllers were not destroyed; firmware was downgraded to a version without monitoring capability and measurements became inaccurate, which made recovery slow.
- Exposure is the first fix. Dragos counted over 46,000 internet-exposed ICS devices speaking Modbus TCP; removing that exposure and segmenting IT from OT addresses most of the documented chain.
Who they are (names and aliases)
FrostyGoop is a tool, not a named group. No source read for this page attributes the attack to a country or a known threat actor. Dragos noted connections from the victim network to Moscow-based IP addresses during the attack but stopped short of attribution, and this page follows that position.
| Name or ID | Source | What it refers to |
|---|---|---|
| FrostyGoop (S1165) | MITRE ATT&CK for ICS | Windows binary written in Go that interacts with ICS equipment over Modbus TCP on port 502, reading and writing holding registers |
| BUSTLEBERM | MITRE ATT&CK alias; Nozomi Networks | Alternative name for the same malware used in other vendor reporting |
| FrostyGoop Incident (C0041) | MITRE ATT&CK | January 2024 campaign against a municipal district heating company in Ukraine |
| Disclosure | Dragos | Found during routine file triage in April 2024, reported to subscribers late May 2024, published 23 July 2024 |
Dragos first thought the samples were test tooling. According to SecurityWeek, it was the Cyber Security Situation Center of the Security Service of Ukraine (SBU) that told Dragos the malware had been used in a disruptive attack, and press reporting places the victim in the Lviv area. For how it fits beside Stuxnet, Industroyer and TRITON, see the ICS Malware History Guide.
Targeting and motivation
- Target: a district heating provider. Dragos assesses with moderate confidence that the malware targeted ENCO heating controllers that had port 502 exposed to the internet.
- Not vendor specific: Dragos stresses that the functionality would work against other devices that accept Modbus TCP, which is why the lesson applies to water, building management and manufacturing sites, not just heating.
- Timing: the disruptive action began on 22 January 2024, during sub-zero temperatures, which maximised civilian impact.
- Motivation: disruption rather than money. No ransom or theft was reported. Sources place it in the context of the war in Ukraine; for the wider pattern of state-linked disruption against civilian infrastructure, see the Cyber Warfare Guide.
- Who should care: any operator with Modbus devices reachable from the internet, from a corporate network or through a vendor connection. Small utilities and building operators with one shared network are the most exposed.
Tactics and techniques (MITRE ATT&CK)
The first four rows are techniques ATT&CK v20 records against S1165. The others describe the wider intrusion as Dragos reported it, mapped by NHI Mgmt Group to the closest verified technique; treat those mappings as analysis, not MITRE's own.
| Stage | Technique | What sources describe | Primary defence |
|---|---|---|---|
| Execution | T1701 Command and Scripting Interpreter (ICS) | Operator-driven tool run from a Windows host | Application control on hosts that can reach OT |
| Collection | T0801 Monitor Process State | Reads holding register values from targets | Baseline which hosts poll which controllers |
| Impair process control | T0836 Modify Parameter | Writes to holding registers to change device operation | Restrict write capability to named engineering hosts |
| Command and control | T0869 Standard Application Layer Protocol; T0885 Commonly Used Port | Plain Modbus TCP on port 502 | Protocol-aware OT monitoring with write alerts |
| Initial access | T1190 Exploit Public-Facing Application | Likely exploitation of an internet-facing router (Nozomi cites MikroTik from the original report) | Patch and harden edge devices; remove management from the internet |
| Persistence | T1505.003 Web Shell | Web shell with tunnelling, mostly reached via Tor | File integrity monitoring on edge and DMZ servers |
| Credential access and movement | T1078 Valid Accounts; T0859 Valid Accounts (ICS) | User credentials taken from the compromised system | MFA and separate OT credentials |
| Command and control | T1572 Protocol Tunneling | L2TP connections to Moscow-based addresses | Egress filtering; alert on new tunnels |
| Initial access (ICS) | T0883 Internet Accessible Device | Controllers with port 502 reachable from outside | Remove OT devices from the public internet |
| Impact | T0829 Loss of View; T0831 Manipulation of Control | Firmware downgraded to remove monitoring; inaccurate readings drove wrong operation | Firmware change control; manual operation fallback |
Hardcoded network routes let attacker-controlled hosts send Modbus commands straight to the controllers. Poor segmentation between the router, management servers and controllers is what turned an edge compromise into a heating outage. For the protocol background, see the Modbus TCP and Modbus Security Guide.
How they abuse identities and credentials
Modbus has no authentication: any host that can open a TCP session on port 502 can issue a write, and the controller trusts it. That makes network position the only "identity" the device checks, so every credential that grants network position into OT is effectively a controller credential.
- Edge device credentials: the router was the foothold. Admin accounts on routers and VPN concentrators that bridge IT and OT need MFA where supported, unique passwords and removal from internet-facing management. See the Edge Device Security Guide.
- Harvested user credentials: Dragos reports the attackers retrieved user credentials after planting the web shell. Shared or reused passwords between corporate and OT systems turn one theft into OT access.
- Default and shared controller passwords: CISA lists changing default passwords immediately as one of five primary OT mitigations. Field devices often ship with shared vendor accounts that nobody owns.
- Remote access accounts: vendor and integrator accounts on remote access paths are long-lived non-human style access. The Colonial Pipeline attack began with a dormant VPN account without MFA; the same class of forgotten account is the obvious route into an OT network. See the OT Remote Access Guide.
Notable incidents
| Date | Event | Source |
|---|---|---|
| Possibly months before January 2024 | Likely exploitation of an external router; web shell deployed and credentials taken | Dragos |
| January 2024 | L2TP connections from the victim network to Moscow-based IP addresses | Dragos |
| 22 January 2024 | Disruptive Modbus activity begins; heating lost for over 600 apartment buildings for almost two days | Dragos; SecurityWeek |
| April 2024 | Dragos finds the binaries in routine triage; no antivirus detections at the time | Dragos |
| 23 July 2024 | Public disclosure of FrostyGoop; MITRE later adds S1165 and C0041 | Dragos; MITRE ATT&CK |
No other confirmed disruptive use of FrostyGoop appears in the sources read. Nozomi published a detection rule and file hashes for defenders; use vendor or ISAC feeds for current indicators rather than copying them from static pages.
Detection and defence
Because the malware uses valid Modbus, detection depends on knowing normal: which hosts talk to which controllers, which function codes they use and when. Dragos frames its advice around the SANS five critical controls for ICS, and CISA's primary OT mitigations cover the same ground.
What to monitor
| Signal | Where to see it | Why it matters |
|---|---|---|
| Modbus write from a host that has never written before | OT network monitoring (passive taps or SPAN) | FrostyGoop's impact came through register writes |
| Modbus traffic crossing the IT/OT boundary or arriving from the internet | Firewall logs at the industrial DMZ | The attack relied on routes from attacker hosts to controllers |
| New Modbus TCP talkers, exceptions or restart and listen-only commands | Protocol-aware IDS; Dragos lists these as detection types | Unusual commands suggest manipulation or reconnaissance |
| Controller firmware version changes | Asset inventory and change records | The attack downgraded firmware to remove monitoring |
| Web shells, new tunnels or Tor connections at the edge | Edge device logs, EDR on DMZ servers, NetFlow | These preceded the OT impact by weeks or months |
| Process values that disagree with physical readings | Operator rounds, independent sensors | Inaccurate measurements were part of the impact |
Phased response plan
- Find exposure first: search your own address space (and an external attack surface service) for port 502 and other OT protocols reachable from the internet; remove them or put them behind a broker. CISA's first mitigation is removing OT connections to the public internet.
- Inventory Modbus devices: list every controller, gateway and HMI that speaks Modbus, with firmware version and owner. See the OT Asset Visibility Guide.
- Baseline in listen-only mode: deploy passive monitoring for two to four weeks, record which hosts read and write, then turn the baseline into alert rules. Passive sensors carry no rollback risk to the process.
- Segment and restrict writes: place controllers in zones with conduits that allow Modbus only from named engineering hosts, with a deny-by-default industrial DMZ. Pilot on one site, keep the previous firewall rule set for rollback, and schedule changes with operations. See the OT Network Segmentation Guide.
- Harden remote and edge access: MFA on every remote path, unique credentials, logging of every session, and timely patching of routers. See the MikroTik RouterOS Hardening Guide where those devices are in use.
- Prepare to run manually: CISA lists practising manual operation as a primary mitigation. Exercise the case where controllers report wrong values, and keep known-good firmware and configuration backups offline. See the OT Incident Response Guide.
Decision guidance
- Small operators (one plant, no OT team): prioritise removing internet exposure, changing default passwords and separating the controller network from office IT with a single managed firewall. A commercial OT monitoring platform can come later.
- Large operators: invest in protocol-aware monitoring across sites with alerts on first-time writes, and treat edge routers as tier-zero assets with the same patch SLA as domain controllers.
- Where Modbus cannot be changed: compensate at the network layer. Modbus/TCP Security (TLS-protected Modbus on port 802) exists but needs device support; most installed controllers will rely on segmentation for years.
Metrics
- Number of OT devices reachable from the internet (target zero, checked monthly).
- Percentage of Modbus devices in the inventory with a known firmware version and owner.
- Percentage of controller zones where only named hosts may issue writes.
- Mean time to alert on a first-time Modbus write in exercises.
- Percentage of remote access paths into OT with MFA and session logging.
- Days since the last manual-operation exercise per site.
Framework mapping
| Control area | NIST SP 800-53 Rev. 5 | NIST CSF 2.0 |
|---|---|---|
| Boundary protection and IT/OT segmentation | SC-7, AC-4 | PR.IR-01 |
| Remote access with MFA | AC-17, IA-2 | PR.AA-03 |
| Asset inventory including firmware | CM-8 | ID.AM-01 |
| Least functionality (writes only from named hosts) | CM-7, AC-6 | PR.PS-01 |
| Edge device patching | SI-2, RA-5 | ID.RA-01, PR.PS-02 |
| OT network monitoring | SI-4 | DE.CM-01 |
| Incident response and manual operation | IR-4, CP-2 | RS.MA-01, RC.RP-01 |
For the OT-specific control baseline, see the NIST SP 800-82 OT Security Guide and the IEC 62443 Guide.
Common failure patterns
- Controllers answering on the internet: port 502 exposed for remote convenience is the direct precondition Dragos describes.
- Flat routing from the edge to field devices: once the router fell, nothing stopped attacker hosts reaching the controllers.
- Relying on antivirus: the binaries were unflagged at discovery; hash-based detection would have missed them.
- No firmware change control: a downgrade that removes monitoring went unnoticed until the impact.
- Treating edge routers as network plumbing: unpatched, internet-managed routers without logging gave months of undetected access.
Practitioner checklist
- Scan for and remove any Modbus TCP (port 502) or other OT protocol exposure to the internet.
- Inventory every Modbus device with firmware version, owner and permitted master hosts.
- Deploy passive OT monitoring and alert on first-time Modbus writes and new talkers.
- Allow Modbus writes into controller zones only from named engineering hosts.
- Patch and harden routers that bridge IT and OT; take their management off the internet.
- Enforce MFA and session logging on all remote access into OT, including vendor accounts.
- Change default and shared controller passwords and keep separate OT credentials.
- Keep known-good firmware and configuration backups offline and track firmware changes.
- Exercise manual operation for the case where controllers report false values.
Standards and references
- MITRE ATT&CK: FrostyGoop (S1165)
- MITRE ATT&CK: FrostyGoop Incident (C0041)
- Dragos: FrostyGoop ICS Malware Intelligence Brief
- CISA: Primary Mitigations to Reduce Cyber Threats to Operational Technology
- Nozomi Networks: Protecting Against FrostyGoop/BUSTLEBERM Malware
- SecurityWeek: FrostyGoop ICS Malware Left Ukrainian City's Residents Without Heating
- NIST: SP 800-53 Rev. 5
- NIST: Cybersecurity Framework 2.0
Related NHI Mgmt Group resources: ICS Malware History Guide · Cyber Warfare Guide · OT Network Security Monitoring Guide · OT Security Guide · Top 10 OT Security Risks