Skip to main content
Back
IntrusionAzure
4 min read

Azure Linux VM seized, likely via a default Cassandra password, to scan other clouds

Victim
An unnamed Microsoft Azure customer
When
Not stated; disclosed November 7, 2018Disclosed Nov 7, 2018
Entry
An Apache Cassandra account with a default password on a Linux VM in Azure, reached by two outside addresses (what Microsoft's analysis suggests).

Microsoft's analysis suggests that two addresses resolving to Romania logged into a customer's Linux VM in Azure with an Apache Cassandra account that kept its default password. From the VM the attacker profiled the host, fetched scanning toolkits and pointed them at two class B ranges, then sprayed over 35,000 SSH credential pairs at ranges of 26 cloud and hosting companies. The customer received multiple Azure Security Center alerts and the activity stopped soon after; no result of the scans or the spray is stated.

What happened

External Parties
Login addresses, Romania
Tool hosting servers
26 providers' address ranges
+ 1 more
Internet
Apache Cassandra
Customer Linux VM in Azure
Linux VM, Cassandra

Replaying the Azure Cassandra VM 2018 intrusion

  1. 1

    Default password

    Date not stated

    Two outside addresses log into the Linux VM, likely with a Cassandra account that kept its default password.

  2. 2

    Profiling and toolkits

    Date not stated

    From the VM the attacker profiles the host, fetches scanning toolkits from internet servers and points them at two class B ranges.

  3. 3

    Password spray on other clouds

    Date not stated

    The VM sprays over 35,000 SSH credential pairs at port 22 across ranges of 26 cloud and hosting companies.

Impact

A customer's Azure Linux VM was taken over and used to scan two class B ranges and to spray SSH passwords at ranges of 26 cloud and hosting companies, until Azure Security Center alerts led to the activity stopping. No result of the scans or the spray is stated.

Outcome

Where Azure network rules fit

If you run a VM like this one, a network security group that admits the database and remote-login ports only from approved ranges would have dropped the logins from the internet, and an outbound rule that denies port 22 to the Internet service tag would have kept the spray inside the VM. Outbound rules that allow only the destinations the VM needs would have bounded the toolkit downloads and the scans, though the source gives no ports for them. The default password still works from any range the rules admit: changing it belongs to the Cassandra installation, which no Azure control reaches.

Could this have been prevented?

In hindsight, not as a whole. 2 controls would have refused a step outright, and 3 controls would have limited one. The counts below are control assessments, not attack stages: several can address the same step. The same replay with the controls in place before the incident shows where each phase would have been refused or bounded.

2
Prevented
3
Contained
0
Not covered

How it could have been prevented

The same replay with the controls in place. Green marks where a step is refused, blue where it is bounded, and red where no control reaches it. Open a row to read the control, why it acts, and its caveat.

External Parties
Login addresses, Romania
Tool hosting servers
26 providers' address ranges
+ 1 more
Internet
Apache Cassandra
Customer Linux VM in Azure
Linux VM, Cassandra

Replaying the Azure Cassandra VM 2018 intrusion

1 · Default password

1 Prevented1 Contained3 on other clouds

2 · Profiling and toolkits

2 Contained6 on other clouds

3 · Password spray on other clouds

1 Prevented3 on other clouds

Key facts

  • Two addresses that resolved to Romania logged into a customer's Linux VM in Azure; Microsoft's analysis suggests an Apache Cassandra account with a default password was used
  • The attacker downloaded and ran a host enumeration script, and Azure Security Center raised an alert for a download from a known malicious source
  • In a second session the attacker fetched a scanning toolkit over a secure FTP connection and pointed it at two class B ranges, just over 65,000 addresses; whether the scans completed is unclear
  • About two hours later the attacker deleted that toolkit and fetched an SSH password spray toolkit from a public website
  • The spray ran over 35,000 username and password pairs against port 22, with target lists of 26 cloud and hosting companies; no successful login is stated
  • The customer received multiple Azure Security Center alerts and the activity stopped soon after; Microsoft published its analysis on November 7, 2018

Sources

Every fact in this breakdown comes from the public reporting below. The replay is a model of that reporting, not a reconstruction of the environment it describes.

  1. 1.
    Leverage Azure Security Center to detect when compromised Linux machines attackMicrosoft Azure Blog · Nov 7, 2018 · Vendor reportInitial access, login sources, host enumeration, toolkits, scanning, password spray, target lists, honeypot contact, detection, end of activity

The estate in the replay is illustrative and drawn at the level of attack stages; the victim's real environment is not public. Verdicts describe what a control does when it is in place before the incident. They are not an assessment of the victim's security program.

More cloud attack breakdowns

All breakdowns