The owners of some 190,000 Docker accounts will need to change their passwords and verify their container images haven't been tampered with as the result of a recent intrusion into a Docker Hub database.
Docker discovered the unauthorized access on April 25. It said it had already notified impacted users about the incident and sent them a password-reset link.
The company said it had also unlinked Docker Hub from GitHub and Bitbucket for those using these external repositories to automatically build — or autobuild — container images. Such users will need to relink their Docker Hub accounts to these repositories in order for autobuild to work properly.
Docker described the intrusion as something that gave attackers a "brief period" of illegal access to a database containing sensitive account information, including usernames and hashed passwords. Also exposed in the breach were tokens that some Docker Hub account owners used to access their repositories on GitHub and Bitbucket. It offered no details on when the breach might have occurred and how it was discovered.
Docker said the 190,000 accounts that had been impacted in the breach represented less than 5% of the overall number of users of its Hub cloud-based container image repository. "No Official Images have been compromised," the company said in a FAQ. Docker pointed to several additional security measures it has in place for protecting its Official Images, including GPG (GNU Privacy Guard) signatures and Notary signing for each image.
Docker Hub is a container image library that developers, software vendors, open source projects, and enterprise software teams use to store and share container images. Many organizations, including large enterprises, use images from the repository to build their containers.
Attacks on the developer pipeline can have a serious impact on application security, says Wei Lien Dang, vice president of product at container security vendor StackRox. "Tainted images can be difficult to detect, and the containers launched from them may even run as expected, except with a malicious process in the background," Lein says.
Since Docker has so far not provided a timeline for the incident, it is unclear how long the attackers might have had access to the compromised accounts, Wei says. To be safe, users need to go back and verify the integrity of any images they might have pushed out over the past several weeks.
Chris Wysopal, CTO of Veracode, says organizations that have been notified should be looking at their logs for signs of unauthorized access, especially if write access was performed. "In this instance, it is critical to review your GitHub logs if you integrated with DockerHub because you will have given DockerHub write access to your repos," he says.
In addition, importing production images to a private registry instead of pulling directly from a public registry can give enterprises more control and separation from events, such as the one involving Docker Hub, Wysopal says. "You can easily go and look at the timeline of the breach and see if you pulled an image during the period the public registry was compromised," he says.
Without any evidence of confirmed malicious tampering, the main takeaway from this breach for organizations is the need to include supply chain attacks in threat modeling exercises, says Tim Erlin, vice president of product management and strategy at Tripwire.
"Organizations that have considered this type of event in their response plan won't be panicking when an incident occurs," Erlin notes. "The key to incident response is to be prepared, and threat modeling allows an organization to identify and prepare for the most relevant threats."
Join Dark Reading LIVE for two cybersecurity summits at Interop 2019. Learn from the industry's most knowledgeable IT security experts. Check out the Interop agenda here.