BlackNevas wipes VMware ESXi estate in 10-day attack

Ransomware operators from the BlackNevas group destroyed an organisation's entire virtualisation estate over ten days, wiping four VMware ESXi hosts, every virtual machine running on them, both vCenter Server Appliances and all connected backup infrastructure, according to an investigation by Kaspersky's Global Emergency Response Team (GERT) reported on 14 November.
What makes the case stand out is what the attackers did not do. They exploited no hypervisor flaw, used no guest-to-host escape and corrupted no hypervisor memory. Every action ran through legitimate VMware management interfaces, so much of the intrusion looked like routine administration. The victim was also running unsupported ESXi builds, which left the environment with fewer defences.
How the attackers got in
The intrusion began with valid accounts. The operators reused administrative credentials taken from systems outside the virtualisation layer to authenticate to vCenter, where their activity appeared operationally normal. From there they connected to the vCenter appliance that managed disaster recovery over SSH as root. Forty-three seconds later they deployed MeshAgent, the endpoint component of the MeshCentral remote-management tool, through a shell script.
The SSH session lasted only two minutes and 26 seconds, and the actors never used SSH on that appliance again. About four hours later they returned through MeshAgent and installed Ligolo-ng version 0.8.2, a reverse tunnelling tool, as a systemd service named vmware-vdma. The service pointed at a binary under /opt/vmware/bin/agent and beaconed to a typosquat domain, api.cloudvflare[.]com, on port 443. Because the attackers stopped using interactive SSH, monitoring that only watched SSH sessions would have missed them.
vCenter turned into a weapon
The most unusual technique was abuse of vCenter Host Profiles. The attackers used the compliance workflow, a tool built to keep ESXi hosts configured to a known good state, to push a change that switched Tech Support Mode SSH from disabled to running. The service reached its running state within 33 seconds, and direct SSH sessions to the ESXi hosts then originated from the trusted vCenter IP address itself.
Defenders can catch this sequence by correlating vCenter events with ESXi logs, since the Host Profile change and the sessions that follow it are recorded. Kaspersky advises treating any Host Profile change that enables SSH outside an approved change window as a serious alert, especially when a new SSH session from the management appliance follows it.
Backups destroyed first
The backup environment was destroyed before production was touched. Days before the final detonation, the attackers disabled antivirus on the backup servers, stopped backup services, encrypted the recovery systems and disrupted administrative access. They also extracted stored credentials from the backup platform, which served as a bridge into other parts of the infrastructure.
That ordering is deliberate. Backup platforms hold service accounts, hypervisor credentials and repository access, making them a convenient pivot point as well as the victim's route back. Removing the restore path first meant that when the virtual machines were wiped, there was nothing clean left to recover from.
The final wipe
The payload arrived over SFTP, and the attackers ran a self-test of about 16 seconds roughly three hours before detonation. The final command wiped datastores to a depth of five, and responders found wipe artefacts across the datastores, with roughly the first two kilobytes of each virtual disk overwritten. That damage destroys the master boot record and the primary GPT header, although the backup GPT may still permit partial recovery in some cases.
On the disaster-recovery host, every virtual machine stopped within 23 seconds, and ESXi logs showed the running virtual machine count dropping to zero. The forward-slash switches used in the wiping command match the Trigona-derived BlackNevas family, and the ransom notes left behind named BlackNevas, a group first observed in November 2024.
What defenders should do
Kaspersky recommends treating vCenter as a Tier 0 asset, giving it dedicated administrative identities separate from enterprise administration, and protecting those accounts with phishing-resistant multi-factor authentication. It also advises keeping the execInstalledOnly runtime option and VIB acceptance levels at their approved baseline, and enabling Lockdown Mode where operations allow.
Logging matters as well. ESXi logs live in a ramdisk that is cleared on reboot, so authentication, shell, management and kernel logs should be forwarded to a remote syslog server. Backup authority should be separated from production Active Directory, and restore tests should be carried out on the assumption that both vCenter and Active Directory are unavailable. Patching alone would not have stopped this attack, because no vulnerability was exploited to reach the hypervisors. The lesson is that the management plane itself needs the same protection as the systems it controls.