Intrusions can take place from both authorized (insiders) and unauthorized (outsiders) users. My personal experience shows that unhappy user can damage the system, especially when they have a shell access. Some users are little smart and removes history file (such as ~/.bash_history) but you can monitor all user executed commands.
It is recommended that you log user activity using process accounting. Process accounting allows you to view every command executed by a user including CPU and memory time. With process accounting sys admin always find out which command executed at what time 🙂
The psacct package contains several utilities for monitoring process activities, including ac, lastcomm, accton and sa.
- The ac command displays statistics about how long users have been logged on.
- The lastcomm command displays information about previous executed commands.
- The accton command turns process accounting on or off.
- The sa command summarizes information about previously executed commmands.
Task: Install psacct or acct package
# yum install psacct
Task: Start psacct/acct service
You need to start psacct service manually. Type the following two commands to create /var/account/pacct file and start services:
# chkconfig psacct on
# /etc/init.d/psacct start
Now let us see how to utilize these utilities to monitor user commands and time.
Task: Display statistics about users’ connect time
ac command prints out a report of connect time in hours based on the logins/logouts. A total is also printed out. If you type ac without any argument it will display total connect time:
Display totals for each day rather than just one big total at the end:
$ ac -d
Nov 1 total 8.65 Nov 2 total 5.70 Nov 3 total 13.43 Nov 4 total 6.24 Nov 5 total 10.70 Nov 6 total 6.70 Nov 7 total 10.30 ..... .. ... Nov 12 total 3.42 Nov 13 total 4.55 Today total 0.52
Display time totals for each user in addition to the usual everything-lumped-into-one value:
$ ac -p
vivek 87.49 root 7.63 total 95.11
Task: find out information about previously executed user commands
Use lastcomm command which print out information about previously executed commands. You can search command using usernames, tty names, or by command names itself.
Display command executed by vivek user:
$ lastcomm vivek
userhelper S X vivek pts/0 0.00 secs Mon Nov 13 23:58 userhelper S vivek pts/0 0.00 secs Mon Nov 13 23:45 rpmq vivek pts/0 0.01 secs Mon Nov 13 23:45 rpmq vivek pts/0 0.00 secs Mon Nov 13 23:45 rpmq vivek pts/0 0.01 secs Mon Nov 13 23:45 gcc vivek pts/0 0.00 secs Mon Nov 13 23:45 which vivek pts/0 0.00 secs Mon Nov 13 23:44 bash F vivek pts/0 0.00 secs Mon Nov 13 23:44 ls vivek pts/0 0.00 secs Mon Nov 13 23:43 rm vivek pts/0 0.00 secs Mon Nov 13 23:43 vi vivek pts/0 0.00 secs Mon Nov 13 23:43 ping S vivek pts/0 0.00 secs Mon Nov 13 23:42 ping S vivek pts/0 0.00 secs Mon Nov 13 23:42 ping S vivek pts/0 0.00 secs Mon Nov 13 23:42 cat vivek pts/0 0.00 secs Mon Nov 13 23:42 netstat vivek pts/0 0.07 secs Mon Nov 13 23:42 su S vivek pts/0 0.00 secs Mon Nov 13 23:38
For each entry the following information is printed. Take example of first output line:
userhelper S X vivek pts/0 0.00 secs Mon Nov 13 23:58
- userhelper is command name of the process
- S and X are flags, as recorded by the system accounting routines. Following is the meaning of each flag:
- S — command executed by super-user
- F — command executed after a fork but without a following exec
- D — command terminated with the generation of a core file
- X — command was terminated with the signal SIGTERM
- vivek the name of the user who ran the process
- prts/0 terminal name
- 0.00 secs – time the process exited
Search the accounting logs by command name:
$ lastcomm rm
$ lastcomm passwd
rm S root pts/0 0.00 secs Tue Nov 14 00:39 rm S root pts/0 0.00 secs Tue Nov 14 00:39 rm S root pts/0 0.00 secs Tue Nov 14 00:38 rm S root pts/0 0.00 secs Tue Nov 14 00:38 rm S root pts/0 0.00 secs Tue Nov 14 00:36 rm S root pts/0 0.00 secs Tue Nov 14 00:36 rm S root pts/0 0.00 secs Tue Nov 14 00:35 rm S root pts/0 0.00 secs Tue Nov 14 00:35 rm vivek pts/0 0.00 secs Tue Nov 14 00:30 rm vivek pts/1 0.00 secs Tue Nov 14 00:30 rm vivek pts/1 0.00 secs Tue Nov 14 00:29 rm vivek pts/1 0.00 secs Tue Nov 14 00:29
Search the accounting logs by terminal name pts/1
$ lastcomm pts/1
Task: summarizes accounting information
Use sa command to print summarizes information about previously executed commands. In addition, it condenses this data into a summary file named savacct which contains the number of times the command was called and the system resources used. The information can also be summarized on a per-user basis; sa will save this iinformation into a file named usracct.
579 222.81re 0.16cp 7220k 4 0.36re 0.12cp 31156k up2date 8 0.02re 0.02cp 16976k rpmq 8 0.01re 0.01cp 2148k netstat 11 0.04re 0.00cp 8463k grep 18 100.71re 0.00cp 11111k ***other* 8 0.00re 0.00cp 14500k troff 5 12.32re 0.00cp 10696k smtpd 2 8.46re 0.00cp 13510k bash 8 9.52re 0.00cp 1018k less
Take example of first line:
4 0.36re 0.12cp 31156k up2date
- 0.36re “real time” in wall clock minutes
- 0.12cp sum of system and user time in cpu minutes
- 31156k cpu-time averaged core usage, in 1k units
- up2date command name
Display output per-user:
# sa -u
root 0.00 cpu 595k mem accton root 0.00 cpu 12488k mem initlog root 0.00 cpu 12488k mem initlog root 0.00 cpu 12482k mem touch root 0.00 cpu 13226k mem psacct root 0.00 cpu 595k mem consoletype root 0.00 cpu 13192k mem psacct * root 0.00 cpu 13226k mem psacct root 0.00 cpu 12492k mem chkconfig postfix 0.02 cpu 10696k mem smtpd vivek 0.00 cpu 19328k mem userhelper vivek 0.00 cpu 13018k mem id vivek 0.00 cpu 13460k mem bash * lighttpd 0.00 cpu 48240k mem php *
Display the number of processes and number of CPU minutes on a per-user basis
# sa -m
667 231.96re 0.17cp 7471k root 544 51.61re 0.16cp 7174k vivek 103 17.43re 0.01cp 8228k postfix 18 162.92re 0.00cp 7529k lighttpd 2 0.00re 0.00cp 48536k
Task: Find out who is eating CPU
By looking at re, k, cp/cpu (see above for output explanation) time you can find out suspicious activity or the name of user/command who is eating up all CPU. An increase in CPU/memory usage (command) is indication of problem.
Please note that above commands and packages also available on other UNIX like oses such as Sun Solaris and *BSD oses.