Techniques
Sample rules
Linux Auditd Service Crash and Restart Burst
- source: splunk
- technicques:
Description
This analytic detects repeated Linux process crash events followed by repeated successful executions of the same executable on the same host within 10 seconds. A rapid crash and restart pattern in a service process can indicate exploitation attempts or unstable application behavior and should be correlated with network and application telemetry.
Detection logic
`linux_auditd`
(
type=ANOM_ABEND
OR
(
type=SYSCALL success="yes"
(
syscall IN ("59", "221", "11")
OR
SYSCALL="execve"
)
)
)
| eval exe=coalesce(exe, executable), event_type=if(type="ANOM_ABEND", "crash", "start")
| eval sig=coalesce(sig, signal)
| where event_type="start" OR tonumber(sig) IN (6, 7, 11)
| sort 0 _time
| streamstats time_window=10s count(eval(event_type="crash")) as crash_count max(eval(if(event_type="crash", _time, null()))) as last_crash_time values(eval(if(event_type="crash", sig, null()))) as crash_signals by host exe
| eval restart_after_crash=if(event_type="start" AND _time>last_crash_time, 1, 0)
| streamstats time_window=10s sum(restart_after_crash) as restart_count by host exe
| where event_type="start" AND crash_count>=2 AND restart_count>=2
| stats min(_time) as firstTime max(_time) as lastTime max(crash_count) as crash_count max(restart_count) as restart_count values(crash_signals) as crash_signals values(comm) as process_name values(uid) as uid values(auid) as auid values(ppid) as parent_pid by host exe
| rename host as dest
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `linux_auditd_service_crash_and_restart_burst_filter`