1AUDITD.CONF(5) System Administration Utilities AUDITD.CONF(5)
2
6 auditd.conf - audit daemon configuration file
7
9 The file /etc/audit/auditd.conf contains configuration information spe‐
10 cific to the audit daemon. Each line should contain one configuration
11 keyword, an equal sign, and then followed by appropriate configuration
12 information. All option names and values are case insensitive. The key‐
13 words recognized are listed and described below. Each line should be
14 limited to 160 characters or the line will be skipped. You may add com‐
15 ments to the file by starting the line with a '#' character.
16 local_events
19 This yes/no keyword specifies whether or not to include local
20 events. Normally you want local events so the default value is
21 yes. Cases where you would set this to no is when you want to
22 aggregate events only from the network. At the moment, this is
23 useful if the audit daemon is running in a container. This op‐
24 tion can only be set once at daemon start up. Reloading the con‐
25 fig file has no effect.
26 log_file
28 This keyword specifies the full path name to the log file where
29 audit records will be stored. It must be a regular file.
30 write_logs
32 This yes/no keyword determines whether or not to write logs to
33 the disk. Normally you want this so the default is yes.
34 log_format
36 The log format describes how the information should be stored on
37 disk. There are 2 options: raw and enriched. If set to RAW, the
38 audit records will be stored in a format exactly as the kernel
39 sends it. The ENRICHED option will resolve all uid, gid,
40 syscall, architecture, and socket address information before
41 writing the event to disk. This aids in making sense of events
42 created on one system but reported/analyzed on another system.
43 The NOLOG option is now deprecated. If you were setting this
44 format, now you should set the write_logs option to no.
45 log_group
47 This keyword specifies the group that is applied to the log
48 file's permissions. The default is root. The group name can be
49 either numeric or spelled out.
50 priority_boost
52 This is a non-negative number that tells the audit daemon how
53 much of a priority boost it should take. The default is 4. No
54 change is 0.
55 flush Valid values are none, incremental, incremental_async, data,
57 and sync. If set to none, no special effort is made to flush
58 the audit records to disk. If set to incremental, Then the freq
59 parameter is used to determine how often an explicit flush to
60 disk is issued. The incremental_async parameter is very much
61 like incremental except the flushing is done asynchronously for
62 higher performance. The data parameter tells the audit daemon to
63 keep the data portion of the disk file sync'd at all times. The
64 sync option tells the audit daemon to keep both the data and
65 meta-data fully sync'd with every write to disk. The default
66 value is incremental_async.
67 freq This is a non-negative number that tells the audit daemon how
69 many records to write before issuing an explicit flush to disk
70 command. This value is only valid when the flush keyword is set
71 to incremental or incremental_async.
72 num_logs
74 This keyword specifies the number of log files to keep if rotate
75 is given as the max_log_file_action. If the number is < 2, logs
76 are not rotated. This number must be 999 or less. The default
77 is 0 - which means no rotation. As you increase the number of
78 log files being rotated, you may need to adjust the kernel back‐
79 log setting upwards since it takes more time to rotate the
80 files. This is typically done in /etc/audit/audit.rules. If log
81 rotation is configured to occur, the daemon will check for ex‐
82 cess logs and remove them in effort to keep disk space avail‐
83 able. The excess log check is only done on startup and when a
84 reconfigure results in a space check.
85 name_format
87 This option controls how computer node names are inserted into
88 the audit event stream. It has the following choices: none,
89 hostname, fqd, numeric, and user. None means that no computer
90 name is inserted into the audit event. hostname is the name re‐
91 turned by the gethostname syscall. The fqd means that it takes
92 the hostname and resolves it with dns for a fully qualified do‐
93 main name of that machine. Numeric is similar to fqd except it
94 resolves the IP address of the machine. In order to use this op‐
95 tion, you might want to test that 'hostname -i' or 'domainname
96 -i' returns a numeric address. Also, this option is not recom‐
97 mended if dhcp is used because you could have different ad‐
98 dresses over time for the same machine. User is an admin de‐
99 fined string from the name option. The default value is none.
100 name This is the admin defined string that identifies the machine if
102 user is given as the name_format option.
103 max_log_file
105 This keyword specifies the maximum file size in megabytes. When
106 this limit is reached, it will trigger a configurable action.
107 The value given must be numeric.
108 max_log_file_action
110 This parameter tells the system what action to take when the
111 system has detected that the max file size limit has been
112 reached. Valid values are ignore, syslog, suspend, rotate and
113 keep_logs. If set to ignore, the audit daemon does nothing.
114 syslog means that it will issue a warning to syslog. suspend
115 will cause the audit daemon to stop writing records to the disk.
116 The daemon will still be alive. The rotate option will cause the
117 audit daemon to rotate the logs. It should be noted that logs
118 with higher numbers are older than logs with lower numbers. This
119 is the same convention used by the logrotate utility. The
120 keep_logs option is similar to rotate except it does not use the
121 num_logs setting. This prevents audit logs from being overwrit‐
122 ten. The effect is that logs accumulate and are not deleted -
123 which will trigger the space_left_action if the volume fills up.
124 This is best used in combination with an external script used to
125 archive logs on a periodic basis.
126 verify_email
128 This option determines if the email address given in ac‐
129 tion_mail_acct is checked to see if the domain name can be re‐
130 solved. This option must be given before action_mail_acct or the
131 default value of yes will be used.
132 action_mail_acct
134 This option should contain a valid email address or alias. The
135 default address is root. If the email address is not local to
136 the machine, you must make sure you have email properly config‐
137 ured on your machine and network. Also, this option requires
138 that /usr/lib/sendmail exists on the machine.
139 space_left
141 If the free space in the filesystem containing log_file drops
142 below this value, the audit daemon takes the action specified by
143 space_left_action. If the value of space_left is specified as a
144 whole number, it is interpreted as an absolute size in megabytes
145 (MiB). If the value is specified as a number between 1 and 99
146 followed by a percentage sign (e.g., 5%), the audit daemon cal‐
147 culates the absolute size in megabytes based on the size of the
148 filesystem containing log_file. (E.g., if the filesystem con‐
149 taining log_file is 2 gigabytes in size, and space_left is set
150 to 25%, then the audit daemon sets space_left to approximately
151 500 megabytes. Note that this calculation is performed when the
152 audit daemon starts, so if you resize the filesystem containing
153 log_file while the audit daemon is running, you should send the
154 audit daemon SIGHUP to re-read the configuration file and recal‐
155 culate the correct percentage.
156 space_left_action
158 This parameter tells the system what action to take when the
159 system has detected that it is starting to get low on disk
160 space. Valid values are ignore, syslog, rotate, email, exec,
161 suspend, single, and halt. If set to ignore, the audit daemon
162 does nothing. syslog means that it will issue a warning to sys‐
163 log. rotate will rotate logs, losing the oldest to free up
164 space. Email means that it will send a warning to the email ac‐
165 count specified in action_mail_acct as well as sending the mes‐
166 sage to syslog. exec /path-to-script will execute the script.
167 You cannot pass parameters to the script. The script is also re‐
168 sponsible for telling the auditd daemon to resume logging once
169 its completed its action. This can be done by adding service au‐
170 ditd resume to the script. suspend will cause the audit daemon
171 to stop writing records to the disk. The daemon will still be
172 alive. The single option will cause the audit daemon to put the
173 computer system in single user mode. The halt option will cause
174 the audit daemon to shutdown the computer system. Except for ro‐
175 tate, it will perform this action just one time.
176 admin_space_left
178 This is a numeric value in megabytes that tells the audit daemon
179 when to perform a configurable action because the system is run‐
180 ning low on disk space. This should be considered the last
181 chance to do something before running out of disk space. The nu‐
182 meric value for this parameter should be lower than the number
183 for space_left. You may also append a percent sign (e.g. 1%) to
184 the number to have the audit daemon calculate the number based
185 on the disk partition size.
186 admin_space_left_action
188 This parameter tells the system what action to take when the
189 system has detected that it is low on disk space. Valid values
190 are ignore, syslog, rotate, email, exec, suspend, single, and
191 halt. If set to ignore, the audit daemon does nothing. Syslog
192 means that it will issue a warning to syslog. rotate will ro‐
193 tate logs, losing the oldest to free up space. Email means that
194 it will send a warning to the email account specified in ac‐
195 tion_mail_acct as well as sending the message to syslog. exec
196 /path-to-script will execute the script. You cannot pass parame‐
197 ters to the script. The script is also responsible for telling
198 the auditd daemon to resume logging once its completed its ac‐
199 tion. This can be done by adding service auditd resume to the
200 script. Suspend will cause the audit daemon to stop writing
201 records to the disk. The daemon will still be alive. The single
202 option will cause the audit daemon to put the computer system in
203 single user mode. The halt option will cause the audit daemon to
204 shutdown the computer system. Except for rotate, it will perform
205 this action just one time.
206 disk_full_action
208 This parameter tells the system what action to take when the
209 system has detected that the partition to which log files are
210 written has become full. Valid values are ignore, syslog, ro‐
211 tate, exec, suspend, single, and halt. If set to ignore, the
212 audit daemon will issue a syslog message but no other action is
213 taken. Syslog means that it will issue a warning to syslog.
214 rotate will rotate logs, losing the oldest to free up space.
215 exec /path-to-script will execute the script. You cannot pass
216 parameters to the script. The script is also responsible for
217 telling the auditd daemon to resume logging g once its completed
218 its action. This can be done by adding service auditd resume to
219 the script. Suspend will cause the audit daemon to stop writing
220 records to the disk. The daemon will still be alive. The single
221 option will cause the audit daemon to put the computer system in
222 single user mode. halt option will cause the audit daemon to
223 shutdown the computer system.
224 disk_error_action
226 This parameter tells the system what action to take whenever
227 there is an error detected when writing audit events to disk or
228 rotating logs. Valid values are ignore, syslog, exec, suspend,
229 single, and halt. If set to ignore, the audit daemon will not
230 take any action. Syslog means that it will issue no more than 5
231 consecutive warnings to syslog. exec /path-to-script will exe‐
232 cute the script. You cannot pass parameters to the script. Sus‐
233 pend will cause the audit daemon to stop writing records to the
234 disk. The daemon will still be alive. The single option will
235 cause the audit daemon to put the computer system in single user
236 mode. halt option will cause the audit daemon to shutdown the
237 computer system.
238 tcp_listen_port
240 This is a numeric value in the range 1..65535 which, if speci‐
241 fied, causes auditd to listen on the corresponding TCP port for
242 audit records from remote systems. The audit daemon may be
243 linked with tcp_wrappers. You may want to control access with an
244 entry in the hosts.allow and deny files. If this is deployed on
245 a systemd based OS, then you may need to adjust the 'After' di‐
246 rective. See the note in the auditd.service file.
247 tcp_listen_queue
249 This is a numeric value which indicates how many pending (re‐
250 quested but unaccepted) connections are allowed. The default is
251 5. Setting this too small may cause connections to be rejected
252 if too many hosts start up at exactly the same time, such as af‐
253 ter a power failure. This setting is only used for aggregating
254 servers. Clients logging to a remote server should keep this
255 commented out.
256 tcp_max_per_addr
258 This is a numeric value which indicates how many concurrent con‐
259 nections from one IP address is allowed. The default is 1 and
260 the maximum is 1024. Setting this too large may allow for a De‐
261 nial of Service attack on the logging server. Also note that the
262 kernel has an internal maximum that will eventually prevent this
263 even if auditd allows it by config. The default should be ade‐
264 quate in most cases unless a custom written recovery script runs
265 to forward unsent events. In this case you would increase the
266 number only large enough to let it in too.
267 use_libwrap
269 This setting determines whether or not to use tcp_wrappers to
270 discern connection attempts that are from allowed machines. Le‐
271 gal values are either yes, or no The default value is yes.
272 tcp_client_ports
274 This parameter may be a single numeric value or two values sepa‐
275 rated by a dash (no spaces allowed). It indicates which client
276 ports are allowed for incoming connections. If not specified,
277 any port is allowed. Allowed values are 1..65535. For example,
278 to require the client use a privileged port, specify 1-1023 for
279 this parameter. You will also need to set the local_port option
280 in the audisp-remote.conf file. Making sure that clients send
281 from a privileged port is a security feature to prevent log in‐
282 jection attacks by untrusted users.
283 tcp_client_max_idle
285 This parameter indicates the number of seconds that a client may
286 be idle (i.e. no data from them at all) before auditd complains.
287 This is used to close inactive connections if the client machine
288 has a problem where it cannot shutdown the connection cleanly.
289 Note that this is a global setting, and must be higher than any
290 individual client heartbeat_timeout setting, preferably by a
291 factor of two. The default is zero, which disables this check.
292 transport
294 If set to TCP, only clear text tcp connections will be used. If
295 set to KRB5, then Kerberos 5 will be used for authentication and
296 encryption. The default value is TCP.
297 enable_krb5
299 This option is deprecated. Use the transport option above in‐
300 stead. If set to "yes", Kerberos 5 will be used for authentica‐
301 tion and encryption. The default is "no". If this option is set
302 to "yes" and it follows the transport option, it will override
303 the transport setting. This would be the normal expected behav‐
304 ior for backwards compatibility.
305 krb5_principal
307 This is the principal for this server. The default is "auditd".
308 Given this default, the server will look for a key named like
309 auditd/hostname@EXAMPLE.COM stored in /etc/audit/audit.key to
310 authenticate itself, where hostname is the canonical name for
311 the server's host, as returned by a DNS lookup of its IP ad‐
312 dress.
313 krb5_key_file
315 Location of the key for this client's principal. Note that the
316 key file must be owned by root and mode 0400. The default is
317 /etc/audit/audit.key
318 distribute_network
320 If set to "yes", network originating events will be distributed
321 to the audit dispatcher for processing. The default is "no".
322 q_depth
324 This is a numeric value that tells how big to make the internal
325 queue of the audit event dispatcher. A bigger queue lets it han‐
326 dle a flood of events better, but could hold events that are not
327 processed when the daemon is terminated. If you get messages in
328 syslog about events getting dropped, increase this value. The
329 default value is 2000.
330 overflow_action
332 This option determines how the daemon should react to overflow‐
333 ing its internal queue. When this happens, it means that more
334 events are being received than it can pass along to child pro‐
335 cesses. This error means that it is going to lose the current
336 event that it's trying to dispatch. This option has the follow‐
337 ing choices: ignore, syslog, suspend, single, and halt. If set
338 to ignore, the audit daemon does nothing. syslog means that it
339 will issue a warning to syslog. suspend will cause the audit
340 daemon to stop sending events to child processes. The daemon
341 will still be alive. The single option will cause the audit dae‐
342 mon to put the computer system in single user mode. halt option
343 will cause the audit daemon to shutdown the computer system.
344 max_restarts
346 This is a non-negative number that tells the audit event dis‐
347 patcher how many times it can try to restart a crashed plugin.
348 The default is 10.
349 plugin_dir
351 This is the location that auditd will use to search for its
352 plugin configuration files.
353 end_of_event_timeout
355 This is a non-negative number of seconds used by the userspace
356 auparse() library routines and the aureport(8) , ausearch(8)
357 utilities to consider an event is complete when parsing an event
358 log stream. For an event stream being processed, if the time of
359 the current event is over end_of_event_timeout seconds old, com‐
360 pared to co-located events, then the event is considered com‐
361 plete. See the NOTES section for more detail.
362
364 In a CAPP environment, the audit trail is considered so important that
365 access to system resources must be denied if an audit trail cannot be
366 created. In this environment, it would be suggested that /var/log/audit
367 be on its own partition. This is to ensure that space detection is ac‐
368 curate and that no other process comes along and consumes part of it.
369 The flush parameter should be set to sync or data.
371 Max_log_file and num_logs need to be adjusted so that you get complete
373 use of your partition. It should be noted that the more files that have
374 to be rotated, the longer it takes to get back to receiving audit
375 events. Max_log_file_action should be set to keep_logs.
376 Space_left should be set to a number that gives the admin enough time
378 to react to any alert message and perform some maintenance to free up
379 disk space. This would typically involve running the aureport -t report
380 and moving the oldest logs to an archive area. The value of space_left
381 is site dependent since the rate at which events are generated varies
382 with each deployment. The space_left_action is recommended to be set to
383 email. If you need something like an snmp trap, you can use the exec
384 option to send one.
385 Admin_space_left should be set to the amount of disk space on the audit
387 partition needed for admin actions to be recorded. Admin_space_left_ac‐
388 tion would be set to single so that use of the machine is restricted to
389 just the console.
390 The disk_full_action is triggered when no more room exists on the par‐
392 tition. All access should be terminated since no more audit capability
393 exists. This can be set to either single or halt.
394 The disk_error_action should be set to syslog, single, or halt depend‐
396 ing on your local policies regarding handling of hardware malfunctions.
397 Specifying a single allowed client port may make it difficult for the
399 client to restart their audit subsystem, as it will be unable to recre‐
400 ate a connection with the same host addresses and ports until the con‐
401 nection closure TIME_WAIT state times out.
402 Auditd events are made up of one or more records. The auditd system
405 cannot guarantee that the set of records that make up an event will oc‐
406 cur atomically, that is the stream will have interleaved records of
407 different events, IE
408 event0_record0
410 event1_record0
411 event2_record0
412 event1_record3
413 event2_record1
414 event1_record4
415 event3_record0
416 The auditd system does not guarantee that the records that make up an
418 event will appear in order. Thus, when processing event streams, we
419 need to maintain a list of events with their own list of records hence
420 List of List (LOL) event processing.
421 When processing an event stream we define the end of an event via
423 record type = AUDIT_EOE (audit end of event type record), or
425 record type = AUDIT_PROCTITLE (we note the AUDIT_PROCTITLE is
426 always the last record), or
427 record type = AUDIT_KERNEL (kernel events are one record
428 events), or
429 record type < AUDIT_FIRST_EVENT (only single record events ap‐
430 pear before this type), or
431 record type >= AUDIT_FIRST_ANOM_MSG (only single record events
432 appear after this type), or
433 record type >= AUDIT_MAC_UNLBL_ALLOW && record type <= AU‐
434 DIT_MAC_CALIPSO_DEL (these are also one record events), or
435 for the stream being processed, the time of the event is over
436 end_of_event_timeout seconds old.
437
440 /etc/audit/auditd.conf
441 Audit daemon configuration file
442
445 auditd(8), audisp-remote.conf(5), auditd-plugins(5).
446
449 Steve Grubb
450Red Hat August 2018 AUDITD.CONF(5)