Security and Compliance
- We practice Server & Site Security
- using only sanitized databases
- taking care to not install restricted access files on development or personal instances outside the project defined security accreditation boundary
- and scrubbing unneeded data from our development systems
- When creating test or exploratory accounts on staging or production systems, we:
- use a user name derived from your name like
- use your
civicactions.comemail address (can add an identifier, e.g.
[email protected]-- everything after the
+is ignored by the mailer)
- use a user name derived from your name like
- We minimize custom code, always preferring to use community maintained modules and contribute patches when needed
- When necessary for new functionality, we strive to create generic modules and contribute them to the parent project
- Custom code must:
- have an associated Jira (or other ticketing system) ticket
- include testing mechanisms, ideally hooked into the continuous integration pipeline
- conform to coding standards (use static code analysis where possible (such as DCQ)
- undergo security peer review
As Drupal Developers
- We follow Drupal coding standards and best practices
- We write Secure Code in Drupal 7
- We understand Security in Drupal 8
- Note that alpha, beta and rc versions are not considered stable and not subject to security team support. In my experience in many cases it is preferable to run a dev than alpha/beta releases where there has been significant number of bug fixes done, and the security profile is identical.
- Each dev release is tied to a specific git commit which is tracked in our repository, and that commit is just as unchangeable as an alpha or beta release (all the latter has is a human readable label for the commit).
- We periodically audit sites to determine if the set of enabled modules are all still in use on the site.
- We ensure that access to documents/sites/dashboards is limited to those that should have access.
- This includes our Google Docs!
- We ensure that users with enhanced privileges (to sites and/or servers)
- must use TFA for authentication/authorization
- are appropriately adjusted upon separation from CivicActions.
Advanced: Connecting to TFA-enabled Sevices/Apps
Some applications and services may need to connect to your Civicactions google account but they might not be able to handle TFA. An example of this would be a personal Gmail account trying to send e-mails through your civicactions' account. For this purpose Google has created something called App Passwords. App Passwords allows you to create a unique password for each of your services/apps. If this password is used while authenticating your service/app to access your CivicActions' account it will bypass TFA.
There are some instructions at https://support.google.com/accounts/answer/185833?hl=en on how to use App Passwords with Google. Several other TFA-enabled services also support app passwords -- see their respecive documentation.
IT: Sharing Service Accounts
- If a service allows individual accounts, use only individual accounts and not shared credentials.
- Prefer services that allow individual accounts, services that allow TFA and secure password policies.
- If a service only allows a single account, have a shared LastPass master account that ideally only 2-3 trusted people have access to. From there share passwords out on an "as needed" basis only, including to individual day-to-day Lastpass accounts for the 2-3 trusted people.
- If the LastPass master account is a paid account it also allows sharing credentials in a way that makes the password harder for the person who you shared it with to recover/view/share (but still allow them to log in with it).
- Shared account passwords should rotate to ensure that only those users needing access continue to have access, revoking individual accounts particularly when people leave.
We use tools to support continuous monitoring for performance and efficiency, and to ensure proper operation and security. These tools include (not an exhaustive list):
- Event and error log capture: auditd (SELinux), fail2ban and AIDE.
- Continuous monitoring dashboards: LogWatch, LogStash/GrayLog, Nagios, Cloudwatch, Pingdom, StatusCake, OpsGenie and New Relic.
- Automated security scanning: OpenSCAP, ansible-role-800-53, and GovReady.
- Every project has an Incident Response Team
- We ensure that at least one member of the Incident Response Team has access to the Internet at all times.
- We train new employees and perform yearly quizzes of employees on the Incident Response procedures.
- Each project can extend or replace the default IRP.