QRadar log sources (WIP)
Reads the log sources configured in IBM QRadar and records, for each host that sends events to it, that the host is covered by the SIEM, through which collectors and protocols, and when QRadar last heard from it. A host QRadar knows is created or enriched as an endpoint, like any other endpoint integration.
| Type name | qradar_log_sources |
| Creates | Endpoints |
| Runs on | A sensor with HTTPS reach to the QRadar console |
| Presence flag | qradar_exists |
| Field prefix | qradar_ |
How it reads QRadar
Every call is a read against the QRadar REST API, with the token sent in the SEC
header. Nothing is changed in QRadar. Each run reads, whole and paged 500 at a time:
| What | QRadar API path |
|---|---|
| Log sources | /api/config/event_sources/log_source_management/log_sources |
| Log source types, groups and extensions | .../log_source_management/log_source_types, /log_source_groups, /log_source_extensions |
| Protocol types | .../log_source_management/protocol_types |
| Event collectors | /api/config/event_sources/event_collectors |
| Host profiles, when Enrich from asset model is on | /api/asset_model/assets and /api/asset_model/properties |
One record per host, not per log source
A single machine usually owns several log sources, for example a Windows Security Event
Log, its WinCollect agent and an IIS source. The sensor folds every log source keyed on
the same host into one record, so each host becomes one change entry. A log source is
keyed on the hostname it reports (its identifier parameter, or the part after @ in
its name), and on its sending IP when it has no hostname. A log source with neither is
skipped.
A sending IP used by more than one log source is treated as a relay (a WinCollect destination or a forwarder), not as the host's own address. It is recorded, but never used to match or address the asset.
QRadar's internal log sources (rule engine, asset profiler, audit) describe QRadar itself and are left out unless Include internal log sources is on.
Asset model enrichment
With Enrich from asset model on, each host is matched to QRadar's own host profile by hostname, FQDN or short name, and by sending IP only when that IP is not a relay. Where several profiles match, the one whose unified name is the host's own, then the least merged one, is used for the operating system, MAC and IP addresses. Every matched profile id is still recorded, so an ambiguous match stays visible. Loopback and link-local addresses are discarded.
Endpoints
Matching
A host is matched to the endpoint it already is by, strongest first:
- Its QRadar hostname, exactly.
- Its short name, in the same DNS domain, on an endpoint whose IP addresses do not contradict it.
- Its full hostname in
names. - Its short name in the same DNS domain, without the address check.
- Its MAC addresses, when the asset model gave any.
- Its own IP addresses, when it has any (never a relay IP), preferring an endpoint whose QRadar hostname does not contradict it.
The domain check reads the endpoint's domains, as a directory integration such as
Active Directory fills it, or the suffix of any name it holds.
An endpoint with no known domain is not excluded by it.
What it writes
| Field | Holds |
|---|---|
qradar_hostname, qradar_sending_ip, qradar_sending_ip_is_shared | How QRadar knows the host, and whether its sending IP is a relay |
qradar_status, qradar_enabled | Health over the enabled log sources, worst first (ERROR, WARN, NA, SUCCESS), or DISABLED when none is enabled |
qradar_log_source_count, qradar_enabled_log_source_count | How many log sources the host has, and how many are enabled |
qradar_log_source_ids, qradar_log_source_names, qradar_log_source_short_names, qradar_log_source_types | The log sources and their types |
qradar_log_source_groups, qradar_protocols, qradar_event_collectors, qradar_extensions | How the events are collected |
qradar_auto_discovered, qradar_wincollect | Whether any source was auto-discovered or is WinCollect managed |
qradar_average_eps | Average events per second, summed over the host's sources |
qradar_last_event_time, qradar_first_creation_date | The newest event from any source, and the oldest source's creation |
qradar_asset_* | The matched host profile: ids, match count and method, domain id, unified name, OS, IPs, MACs, products, vulnerability count and risk score |
qradar_log_source_details | The full log source records, as JSON |
It also writes the shared fields names (short name), fqdn and domains (added to
what the endpoint holds), mac (added), os, os_vendor, private_ip and
updated_at. type is set only when the operating system says what the device is.
Without asset model enrichment there is no OS, MAC or profile data, and private_ip
comes from the sending IP only when that is not a relay.
The newest event time is the endpoint's last-seen time and is decommissioning evidence. The oldest log source's creation date is its first-seen time.
A host whose newest event is older than Ignore of host is offline (180 days by default) is a stale log source definition, not a live machine, and is skipped so it does not recreate decommissioned assets. A host that has never sent an event is not skipped.
Requirements
Network access
| From | To | Protocol | Port |
|---|---|---|---|
| The sensor running the integration | The QRadar console named in URL | HTTPS over TCP | 443, unless the URL names another |
If the console presents a certificate the sensor host does not trust, install its CA on the host in preference to enabling Ignore SSL: the token is sent on every request.
Access in QRadar
| Need | Why |
|---|---|
| An authorised service token | Sent as the SEC header on every call |
| The Log Source Management capability | Every run reads the log source management and event collector paths above in full; the connection test reads one log source to prove it |
| The Assets capability | Only when Enrich from asset model is on; the connection test then also reads one asset to prove it |
Give the integration a token of its own, under a security profile that covers the hosts it should see, rather than an administrator's.
Confirm the least-privileged QRadar user role that holds these capabilities, and whether the event collectors path needs any capability beyond Log Source Management.
In this deployment
| Need | Why |
|---|---|
| A registered, healthy sensor with reach to the console | The integration runs on a sensor, on the platform's schedule, see Sensors |
Configuration
| Field | Required | Notes |
|---|---|---|
| Name | Yes | Label for this instance |
| URL | Yes | QRadar console, e.g. https://qradar.example.com |
| Token | Yes | Authorised service token. Write-only: once saved it is never shown again, only replaced, and is sent only to the sensor |
| Cookies | No | Extra cookies added to requests. Write-only like Token, and can be removed |
| Proxy settings | No | http://host:port |
| Ignore SSL | No, on by default | Skips certificate validation |
| Enrich from asset model | No, on by default | Reads OS, MAC and IP details from the QRadar asset model |
| Include internal log sources | No, off by default | Also imports QRadar's own log sources as an asset |
| Sensor | Yes | Sensor that will run the integration |
Its scan template adds nothing to the shared model. Ignore of host is offline defaults to 180 days and decides which silent hosts are dropped, as above.
See also
- Integrations overview: the shared configuration model
- Sensors: the component that runs the integration
- Assets: which source asserted each value on an endpoint
- Active Directory (WIP): fills the
domainsthe name rules check - CrowdStrike: agent-side view of the same endpoints