Repository navigation
Conversation
Adds docs/contributing-security-model.md, describing how Malcolm's components trust one another, where its security boundaries are, and which configuration choices move them. Covers: - the internal container network as a single trust domain, and why internal services trust their reverse proxy - what the installer publishes to the outside, under which "Expose Malcolm Service Ports" selections, and that exposed services bind 0.0.0.0 rather than loopback - the aggregator/sensor reachback enclave - the host boundary: the PUSER_PRIV_DROP privilege-drop model, the difference between dropping at the entrypoint and dropping per supervisord program, and which processes legitimately remain root - nginx as the authentication layer, the identity headers backends consume, what RBAC does and does not separate, and the RBAC URI normalization problem - data ingestion and expensive queries as trusted user capabilities - configuration placeholders substituted at container startup, and build-time credential handling - deployment choices that move the boundaries, and what is out of scope, including the pinning and checksum verification the build already does Also adds SECURITY.md covering vulnerability reporting, what to expect after a report, and CVE assignment through CISA's VINCE system, and links the new document from the contributor guide and documentation index. Fixes cisagov#1279 Signed-off-by: Seth Grover <seth.d.grover@gmail.com>
Collaborator
|
@berlikm thanks for the contribution! I've just glanced at it and will get to a more in-depth triage/review hopefully in the not-to-distance future. At the moment we're a little bit resource-constrained and swimming in PRs, so please have patience with us and we'll try to get this slated for either this month's or next month's release. Again, we appreciate you pitching in! |
Preserve upstream TLS verification and same-origin API URL handling alongside endpoint reconciliation and host-mask policy.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Passive NetBox discovery can create a duplicate device and
/32when a host already exists as an unassigned IP record, ignore existing native MAC objects, and post the read-only interfacemac_addressfield used before NetBox 4.2. An opt-in policy is also needed to use known network-prefix lengths for newly created host addresses.The change guards existing host IPs before creating candidates, resolves unique native MAC identities in a matching prefix/VLAN/site context, and uses the current MAC-address API for eligible auto-created device/VM interfaces. Imported assignments and existing primary MACs are preserved. Contextual cache keys and process-local site/host locks prevent lookup leakage and concurrent duplicate creation.
NETBOX_AUTO_POPULATE_INHERIT_PREFIX_LENGTHdefaults tofalse. When enabled, new host IPs inherit the most specific unique containing IPAM prefix's length only where site and VRF context are consistent. Site resolution supports direct site/location scope or a prefix's VLAN/site-scoped VLAN group. Container prefixes, missing/ambiguous scope and unresolved VRFs retain/32or/128. Explicit masks and existing IP records are preserved. This is inventory policy, not a mask measured from packet headers.Mask policy and MAC ownership use separate rules. A routed endpoint may be observed on another VLAN, and a first-observation log such as Zeek
known_hostsmay omit VLAN metadata entirely. Both can still use an authoritative scoped IPAM mask. For example, an untagged first event for10.10.0.7, followed by traffic on a different capture VLAN, can create one10.10.0.7/24from the known prefix. Matching capture/native VLAN context remains mandatory for MAC lookup/assignment, so this does not assign gateway MACs to endpoints. The option is wired into the enrichment pipeline and documented in the configuration reference and example environment file.Validation:
/24without an observed VLAN, accepted the same scoped mask on a routed observation, rejected the gateway MAC, verified primary IP and removed all temporary fixtures.Existing duplicate records and historical assignments require separate reconciliation. MAC attribution remains limited by local observation coverage. Synchronization is local to one Logstash process rather than a distributed lock across multiple Malcolm deployments.