Skip to navigation

Threat reportMalwareTL-2026-1745

Tengu Botnet Reboots Compromised Linux Devices When Defenders Kill Its Process

highACTIVE

Tengu is a Mirai-derived Linux/IoT botnet, documented by Nozomi Networks Labs, that compromises internet-exposed embedded devices across six CPU architectures via Telnet credential brute force. It survives takedown attempts through a 60-second guardian-process respawn, fake systemd/init/RC/shell-startup persistence, binary immutability marking, and hardware-watchdog abuse that force-reboots the device if its process is killed, while offering 25 DDoS methods, SOCKS5 proxying, remote shell execution, and a custom ChaCha20/Poly1305-like encrypted C2 channel.

Key facts

IdentifiersTL-2026-1745
SeverityHIGH
Affected productsMultiple / unspecified OEM embedded-Linux device manufacturers Internet-exposed embedded Linux IoT devices (routers, IP cameras, DVRs, and Android TV boxes) running with Telnet or similar remote administration enabled
Weaknesses (CWE)CWE-521, CWE-306
MalwareMirai, tengu
ToolsCustom ChaCha20/Poly1305-like encrypted C2 protocol, SOCKS5 proxy module
MITRE ATT&CK techniquesT1016 System Network Configuration Discovery, T1027 Obfuscated Files or Information, T1036 Masquerading, T1037 Boot or Logon Initialization Scripts, T1053 Scheduled Task/Job, T1059 Command and Scripting Interpreter, T1070 Indicator Removal, T1071 Application Layer Protocol, T1082 System Information Discovery, T1090 Proxy, T1102 Web Service, T1105 Ingress Tool Transfer and 13 more in the attack flow
MotivationUNKNOWN
Target sectorsiotconsumerdevices, smallofficehomeoffice, embeddedsystems, generalenterpriseiot
Target regionsGlobal
Detection rules9 (1 shown free)
Sources6 cited sources

Detection coverage

As of 2026-07-28, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-1745 across Splunk SPL, Microsoft KQL and Sigma, covering 20 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: Encrypted C2 Beacon and XOR-Obfuscated C2 Address Handling to Tengu Infrastructure (64.89.163.8:9931)

Detects outbound TCP connections from an embedded/IoT device to the known Tengu C2 server 64.89.163.8 on port 9931 — the registration/heartbeat/command channel where plaintext registration and heartbeat traffic transitions into a custom ChaCha20/Poly1305-like encrypted command channel — combined with periodic beacon-interval/jitter timing consistent with Tengu's heartbeat framing.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1573.001, T1071, T1027
  • Data sources: network flow/netflow logs, firewall connection logs, DeviceNetworkEvents
  • Severity: critical
  • Confidence: high

Splunk SPL

index=network sourcetype IN ("stream:tcp","pan:traffic","firewall","netflow")
dest_ip="64.89.163.8" dest_port=9931
| bin _time span=5m
| stats count as conn_count, values(src_ip) as devices, min(_time) as first_seen, max(_time) as last_seen by dest_ip, dest_port
| eval avg_interval=(last_seen-first_seen)/conn_count
| eval threat="Tengu encrypted C2 beacon to known infrastructure (T1573.001, T1071, T1027)"
| table dest_ip, dest_port, devices, conn_count, avg_interval, threat

Microsoft KQL

DeviceNetworkEvents
| where Timestamp > ago(24h)
| where RemoteIP == "64.89.163.8" and RemotePort == 9931
| summarize ConnCount = count(), Devices = make_set(DeviceName), FirstSeen = min(Timestamp), LastSeen = max(Timestamp) by RemoteIP, RemotePort
| extend AvgIntervalSeconds = datetime_diff('second', LastSeen, FirstSeen) / ConnCount
| project RemoteIP, RemotePort, ConnCount, Devices, FirstSeen, LastSeen, AvgIntervalSeconds

Sigma

title: Tengu Botnet - Encrypted C2 Beacon to Known Infrastructure (64.89.163.8:9931)
id: 8c1e5f47-2a9d-4b6c-9e3a-5d7f1c8b4a20
status: experimental
description: Detects outbound TCP connections to the confirmed Tengu botnet C2 server on its documented registration/heartbeat/command port, which carries plaintext registration traffic followed by a custom ChaCha20/Poly1305-like encrypted command channel.
references:
    - https://www.nozominetworks.com/labs
    - https://urlhaus.abuse.ch/host/64.89.163.8/
author: Threadlinqs Team
date: 2026-07-28
tags:
    - attack.command-and-control
    - attack.t1573.001
    - attack.t1071
    - attack.t1027
    - attack.tool.tengu
    - attack.tool.mirai
logsource:
    category: network_connection
    product: linux
detection:
    selection:
        DestinationIp: '64.89.163.8'
        DestinationPort: 9931
    condition: selection
falsepositives:
    - None expected — this IP/port pair is a dedicated, community- (TL_OSINT_Scan/tweetfeed.live, reporter 'deobfuscately', AS401626) and vendor-corroborated Tengu C2 endpoint with no known legitimate service
level: critical

Known false positives:

  • None expected against this specific C2 IP:port pair; if the IP is reallocated by the hosting/ASN provider in the future, refresh the IOC

Tuning notes: This detection is IOC-driven and will go stale if Tengu rotates C2 infrastructure — pair with the process-masquerading and dropper-execution detections (behavioral, not IOC-bound) for durable coverage; TL_OSINT_Scan/tweetfeed.live independently corroborates 64.89.163.8 as sighted infrastructure (first seen 2026-06-21, predating this report), reinforcing confidence but remaining community/OSINT-tier corroboration, not ground truth.

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • Multi-Architecture ELF Dropper Fetch-and-Execute via Shell Script (Tengu Stage-2 Delivery) · unlock with TL Creds
  • Process Masquerading as [indicator] or [indicator] with Anomalous Lineage · unlock with TL Creds
  • SOCKS5 Proxy-Relay Abuse and IPFS Payload-Gateway Retrieval from Compromised IoT Devices · unlock with TL Creds
  • Successful Telnet Authentication Followed by Outbound Fetch to Known Tengu C2/IPFS Host · unlock with TL Creds
  • Fake Systemd/RC/Cron Persistence Installation Combined with ELFOOD Reboot-Utility Corruption and Watchdog Abuse · unlock with TL Creds
  • Dormant DGA Domain-Resolution Attempts and XOR-Obfuscated C2 Fallback Activation · unlock with TL Creds
  • Telnet Brute-Force Success Following Rapid Failed Login Attempts (Tengu Initial Access) · unlock with TL Creds
  • Guardian-Process Respawn Loop and chattr Immutability Marking (Tengu Anti-Remediation) · unlock with TL Creds

Indicators of compromise

As of 2026-07-28, Threadlinqs Intelligence tracks 20 indicators of compromise for TL-2026-1745.

Sample indicators (1 of 20)

  • IP address: 64.89.163[.]8

Defanged, TLP:CLEAR. Full set of 20 indicators: unlock with TL Creds.

20 indicators of compromise · unlock with TL Creds. Compare plans

Community OSINT corroboration

2 of this threat's indicators have also been reported by the open-source security community, which observed at least one of them before this report was published. Community sightings are unverified and are kept separate from Threadlinqs' curated indicators. Indicator values, reporters and campaign linkage are available to authenticated Red-tier users.

Technical analysis

Tengu is a newly documented Mirai-lineage botnet targeting internet-facing embedded Linux systems — routers, cameras, DVRs, and other low-maintenance devices — that expose Telnet or similar remote administration services. Nozomi Networks Labs published the first technical analysis on 2026-07-27, and the finding was corroborated the following day by The Hacker News, Cyber Security News, Cyberpress, and Cryptika Cybersecurity.

Initial access is achieved through Telnet credential brute force against honeypot and internet-exposed devices. A shell-script dropper subsequently pulls an architecture-specific binary over HTTP, with samples confirmed for i386, amd64, MIPS, ARM, PowerPC, and m68k — giving Tengu one of the broadest hardware footprints observed in a current Mirai derivative.

What separates Tengu from generic Mirai clones is its anti-remediation depth. The primary malware process forks a detached guardian that checks the main process every 60 seconds and relaunches the installed binary if it is stopped. In parallel, a background worker masquerading as the kernel thread "[kworker/0:0]" reopens the hardware watchdog device where available, arms it with an approximately 30-second timeout, and issues keepalive signals only while the main malware process remains alive — meaning that if a defender kills the process without disabling the watchdog, the device is forced into a hardware reboot, destroying volatile forensic evidence and undoing in-memory remediation. Tengu additionally marks its installed binary immutable, installs fake systemd service units, alters init/RC scripts and shell startup files (cron-based persistence is present but described as unfinished/broken), and overwrites the ELF headers of a hardcoded list of reboot/shutdown utility binaries with the string "ELFOOD" — corrupting the very tools an administrator would use to cleanly power-cycle the device.

The malware also runs largely fileless: it uses memfd_create() to build an in-memory file named "systemd-journal" (falling back to /dev/shm/.journal if memfd is unavailable), deletes the filesystem-visible entry while keeping the file descriptor open, and re-executes itself via execve — leaving little on-disk footprint. The main process additionally rewrites its displayed command-line name to /usr/lib/systemd/systemd-journald to blend into normal process listings, and performs self-integrity checks by reading /proc memory-mapping information, computing a baseline SHA-256 value over part of its own code, and repeatedly comparing it (alongside monitoring for unexpectedly writable memory mappings) to detect tampering, patching, or analysis.

Command and control runs to 64.89.163.8 over TCP/9931. Registration, heartbeat, and command-output traffic are sent in plaintext, but server-issued commands and binary updates are protected with a custom ChaCha20/Poly1305-like authenticated encryption scheme. The C2 IP is XOR-obfuscated inside the binary, and dormant domain-generation-algorithm (DGA) logic is present but not observed in active use, suggesting a fallback capability the operator has not yet activated. The same host also serves an IPFS gateway (port 8080) used to retrieve additional ELF or APK payloads, including an Android APK whose delivery path suggests targeting of poorly secured Android TV boxes, though no confirmed Android infections have been documented.

Operational capabilities include 25 distinct DDoS methods, built-in SOCKS5 proxying (enabling proxy resale/relay abuse), arbitrary remote shell command execution, system and network reconnaissance data collection, self-updating, and the ability to detect and remove competing malware from a compromised device — a common Mirai-family trait aimed at monopolizing device resources.

URLhaus has tracked 17 malware URLs at 64.89.163.8 since 2026-06-17 (most recently 2026-07-07), comprising shell-script droppers, multiple ELF binaries tagged as Mirai, and one APK — corroborating that the C2 host doubles as an active malware distribution point. No CVE applies: this is a configuration/credential-exposure threat (exposed Telnet plus default or weak credentials) rather than a specific software vulnerability, so remediation is centered on reducing exposure and hardening credentials rather than patching.

MITRE ATT&CK techniques by tactic:

Discovery

T1016 System Network Configuration Discovery; T1082 System Information Discovery

Defense Evasion

T1027 Obfuscated Files or Information; T1036 Masquerading; T1070 Indicator Removal; T1497 Virtualization/Sandbox Evasion; T1620 Reflective Code Loading

Persistence

T1037 Boot or Logon Initialization Scripts; T1053 Scheduled Task/Job; T1543 Create or Modify System Process; T1546 Event Triggered Execution

Execution

T1059 Command and Scripting Interpreter

Command and Control

T1071 Application Layer Protocol; T1090 Proxy; T1102 Web Service; T1105 Ingress Tool Transfer; T1568 Dynamic Resolution; T1573 Encrypted Channel

credential-access

T1110 Brute Force

Initial Access

T1133 External Remote Services

defense-impairment

T1222 File and Directory Permissions Modification; T1685 Disable or Modify Tools

Impact

T1489 Service Stop; T1498 Network Denial of Service; T1529 System Shutdown/Reboot

Remediation

Patches

  • No vendor patch applies — Tengu propagates via exposed Telnet and weak/default credentials rather than a specific software vulnerability; remediation is configuration-based

Immediate actions

  • Disable Telnet and any other unauthenticated or default-credential remote administration service on internet-facing IoT/embedded Linux devices
  • Block inbound and outbound traffic to 64.89.163.8 on TCP/9931 (C2) and TCP/8080 (IPFS gateway) at the network perimeter
  • Replace all default/factory-set device credentials with strong, unique passwords
  • Flag and forensically preserve (before rebooting) any device running a process masquerading as systemd-journald with anomalous behavior, or a [kworker/0:0]-named process holding active network sockets

Workarounds

  • Before returning a cleaned device to service, inspect systemd service units, init/RC scripts, shell startup files, and cron entries for Tengu-installed persistence artifacts
  • Because killing the malware process can trigger a hardware-watchdog-forced reboot, plan remediation as a full firmware re-flash or factory reset rather than in-place process termination

Longer-term hardening

  • Segment IoT and embedded devices onto isolated network zones separate from critical enterprise infrastructure
  • Deploy network-based monitoring for anomalous outbound connections, DDoS-pattern traffic, and SOCKS5-proxy-style relay traffic originating from IoT device subnets
  • Maintain an active firmware/patch update program for internet-facing embedded devices
  • Where feasible, deploy Linux-capable EDR/behavioral monitoring able to flag memfd_create-based fileless execution and process-name masquerading

Affected and fixed versions

  • Multiple / unspecified OEM embedded-Linux device manufacturers — Internet-exposed embedded Linux IoT devices (routers, IP cameras, DVRs, and Android TV boxes) running with Telnet or similar remote administration enabled
    Vulnerable versions: i386 builds; amd64 builds; MIPS builds; ARM builds; PowerPC builds; m68k builds
    Fixed in: Not applicable — mitigated via configuration (disable Telnet, enforce strong credentials), not a vendor patch

Timeline

  • URLhaus records the first malicious activity at the Tengu C2 host locked-ioc-3, marking the earliest documented infrastructure activity tied to this campaign.
  • Most recent URLhaus submission tied to locked-ioc-3, bringing the tracked total to 17 malware URLs (shell-script droppers, Mirai-tagged ELF binaries, and one APK) hosted at the address.
  • Nozomi Networks Labs completes technical analysis of the Tengu botnet, documenting its guardian-process respawn, fake-systemd persistence, and hardware-watchdog-driven forced-reboot anti-remediation behavior.
  • Nozomi Networks Labs publishes a SHA-256 hash for its analyzed Tengu sample; as of this date no URLhaus-submitted binary at locked-ioc-3 matches the published hash, suggesting active binary rotation or a build distinct from the URLhaus-captured samples.
  • All 17 URLhaus-tracked malware URLs previously hosted at locked-ioc-3 are confirmed offline as of publication, though the underlying C2 IP itself is not confirmed decommissioned and may resume serving payloads.
  • Cryptika Cybersecurity publishes defensive guidance recommending Telnet exposure reduction, credential hardening, and inspection of persistence locations before returning affected devices to service.
  • Cyberpress publishes a deep-dive on Tengu's memfd_create-based fileless execution and its systemd-journald/kworker process masquerading.
  • Cyber Security News corroborates the Nozomi findings, emphasizing the SHA-256 self-integrity check and the reboot-on-kill defense mechanism.
  • The Hacker News publishes the first public report on Tengu, detailing its six-architecture reach (i386, amd64, MIPS, ARM, PowerPC, m68k), 25 DDoS methods, SOCKS5 proxying, and ChaCha20/Poly1305-like C2 encryption.

Sources cited

Previous and next in malware

Drafted by the Threadlinqs research pipeline from 6 sources; published ; last updated —.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats