Log Forwarding & SIEM
JumpServer can forward application logs to an external Syslog server, so that the records survive the loss of a JumpServer host and can be collected, searched, and correlated in a SIEM. The forwarded messages carry a JSON payload that includes user, address, organization, and operation fields, which makes them suitable for SIEM parsing.
Overview
The flow has three parts:
The forwarded records include login, file transfer, operation, password change, session, and command logs; see Log payload format.
- Every component writes its logs under
VOLUME_DIR,/data/jumpserverby default. The per-component directories are listed in Troubleshooting. - Forwarding is enabled by deployment settings in
/opt/jumpserver/config/config.txt:SYSLOG_ENABLE,SYSLOG_ADDR,SYSLOG_FACILITY, andSYSLOG_SOCKTYPE. The change takes effect after a service restart. See Configuration and Integrations. - The collector receives an application log stream, not the audit database. Login, operation, and session records remain in JumpServer and are reviewed in Audits; the Syslog stream is a copy of selected application logs, not a replacement for the audit records.
The Syslog target is deployment configuration, so one target applies to the
whole host. Transport security (TLS/RELP) and the accepted values of
SYSLOG_SOCKTYPE are not documented; read the sections below before you plan
a SIEM integration.
Syslog settings
The settings are deployment configuration in
/opt/jumpserver/config/config.txt:
Pending confirmation: the accepted values of SYSLOG_SOCKTYPE, whether
transport security (TLS/RELP) is supported, and whether every component honors
the deployment Syslog settings. Do not assume that the stream is encrypted.
Enable forwarding
Back up the deployment configuration file.
Set the Syslog values in config.txt. The address includes the port, and the
facility must match the collector configuration.
Restart the services so the containers read the new values.
Send a test message and confirm that the collector receives it. See Retention and validation.
./jmsctl.sh restart interrupts active sessions. Apply the change in a
maintenance window and notify users first.
Collector preparation
A standard Linux Syslog server can receive the messages. The example below uses
rsyslog with UDP on port 514 and stores the messages in a dedicated file.
Enable UDP reception and route the JumpServer facility in /etc/rsyslog.conf:
Restart the collector service.
Send a test message from the JumpServer host, replacing the address with your collector.
Confirm that the message appears in the collector output file before you continue. If it does not, check the UDP listener and the firewall between the hosts.
Log payload format
Each forwarded message is a Syslog line followed by a structured JSON payload.
The payload starts after the jumpserver: <log_name>- marker:
The log name identifies the record type:
session_command_log includes the command output. Treat the Syslog stream as
sensitive data: restrict collector and SIEM access, and apply the same
retention and masking policy that you apply to session records.
The payload is JSON, not CEF or LEEF. Parse it after the jumpserver: <log_name>- marker, and keep the Syslog header for the timestamp, host, and
facility. Field sets can differ between record types; build parsers per log
name.
Coverage matrix
The table lists where each source writes locally and whether this documentation states that it is forwarded. The deployment setting is documented at the host level, not per component.
Pending confirmation: whether the component runtime logs (KoKo, Chen, Razor,
Magnus, Web) and the built-in PostgreSQL and Redis containers are also forwarded
by the deployment settings. System Settings > Components > Log shows
component log streaming when log collection is enabled, but its relationship to
the SYSLOG_* settings is not documented.
Transport security
The documented example uses UDP on port 514, and SYSLOG_SOCKTYPE selects the
transport. TLS and RELP are not documented, and the message body is plain JSON.
If encrypted transport is a requirement, raise it with the vendor before the
integration is planned.
Do not expose the collector to untrusted networks, and do not put a TLS terminator in front of a plain Syslog stream and assume that the forwarded messages are authenticated.
Multi-node aggregation
Each JumpServer node has its own deployment configuration. To aggregate a multi-node deployment:
- Point every node's
SYSLOG_ADDRat the same collector so that all application logs arrive in one place. - Distinguish the nodes at the collector by source address, because all nodes can use the same facility and format.
- Configure NTP on every node with
./jmsctl.sh config ntpso that timestamps from different nodes line up in the SIEM. See Command-Line Tools. - Restart the services on each node after editing its configuration, following the order described in Upgrade & Maintenance.
Pending confirmation: a shared or high-availability recommendation for the
Syslog configuration itself. This documentation does not state whether nodes
must share the same SYSLOG_* values or can use different ones; the guidance
above is the general approach for per-host configuration.
Retention and validation
Retention is a collector-side decision. JumpServer documents retention for its own records, not for forwarded logs, and the regular clean-up under System Settings > System Tasks > Regular clean-up applies to platform records, not to the Syslog stream. Set the retention period on the collector to match the audit policy.
Validate delivery in the order below:
Confirm that the collector is reachable from the JumpServer host on the
address and port in SYSLOG_ADDR, and that the transport matches your
SYSLOG_SOCKTYPE. Confirm that the outbound direction is allowed; see
Network Ports.
Send a test message from the host. The example uses UDP and the facility configured above.
If the collector does not receive the test message, capture the traffic on the host to confirm that the messages leave the interface. Use the port you configured.
Trigger a real JumpServer event, such as a sign-in, an operation, or a session,
and confirm that the corresponding jumpserver: <log_name>- line arrives with
its JSON payload.
Only after both messages arrive, write the SIEM parsing and alerting rules against the captured payloads, one parser per log name.
logger and tcpdump are host tools, not JumpServer commands. Skip a step
when the tool is not installed on the host, and use the collector or a network
capture point instead.
Troubleshooting
For the general diagnostic flow, see Troubleshooting. For the ports the platform itself needs, see Network Ports.