Overview

In this assignment you will use BASH and various system utilities to construct a system

monitor that periodically reports and alerts on system activities such as Network Usage

$($number of connections, data sent, data received$),$ Disc I.O$.,$ Running Processes $($including

their CPU, and Memory consumption$),$ System CPU and Memory consumption, the system

temperature, and system load.

To be successful you will need to create several bash functions, utilize and parse the output of

multiple commands $($this may require using parsing commands$).$ A helpful list of commands

to research:

$$ top

$$ vmstat

$$ iostat

$$ netstat

$$ ifstat

$$ uptime

$$ ps

$$ df

$$ wc

$$ grep

$$ sed

$$ cut

$$ awk

$$ date

$$ cat

$$ sort

$$ head

If you need to find a way to get data and don$$t know the command apropos can help. For

some items you may need to get information from the $/$sys system

$($i$.$e$./$sys$/$class$/$thermal$/$thermal zone$0).$ You can explore the various classes there for if

there are ones of use.

$2$

System Monitor $[80$ Points$]$

Your system monitor will be involved with the following command format:

$./$system$\_$monitor.sh $1060./$sys$\_$monitor.log "cpu$>80$ mem$>90$ disk$>75"$ eno$0\backslash$

sda $/$sys$/$class$/$thermal$/$thermal$\_$zone$0/$temp

$($line break for clarity$)$ where the arguments are as follows:

$$ Interval $($e$.$g$.10$ seconds how often to check all parameters$)$

$$ Duration $($e$.$g$.60$ seconds how long to run for$)$

$$ Logfile $($e$.$g$.60./$sys monitor.log where to log the data$)$

$$ Alerts $($e$.$g$.$cpu$80$ mem$90$ disk$75$ What tests should trigger an alert$)$

$$ Network Interface $($e$.$g$.$ eno$0$ what network interface to monitor$)$

$$ Drive $($e$.$g$.$ sda what disc to monitor$)$

$$ Temperature Node $($e$.$g$./$sys$/$class$/$thermal$/$thermal zone$0/$temp what tempera$-$

ture node to monitor$)$

You should always monitor the following parameters:

$$ cpu Current Total CPU usage, unit: $\%$

$$ mem Current Total MEM usage, unit: $\%$

$$ net tx Current Total Network Transmit, unit: packets

$$ net rx Current Total Network Receive, unit: packets

$$ net con Current Total Network Connections, unit: none

$$ disk Current Total Disk usage, unit: kB

$$ disk W Current Total Disk write, unit: kB

$$ disk R Current Total Disk read, unit: kB

$$ proc Current Total Number of Processes $($all users$),$ unit: none

$$ proc cpu Current Process $($name$)$ with highest cpu usage, unit: none

$$ proc mem Current Process $($name$)$ with highest memory usage, unit: none

$$ temp Current system temperature, unit: C

$$ load Current system $1$ minute load average, unit: none

$3$

Logging

Your monitor should output these parameters to a log file at each interval on a single line

with no units separated by a space preceeded by a timestamp in the format YYYYMMDD$-$HHMM:

$20240707-1435$ cpu$=50$ mem$=75,$ net$\_$tx$=750$ net$\_$rx$=500$ net$\_$con$=500$ disk$=500532\backslash$

disk$\_$W$=44001039$ disc$\_$R$=5181087$ proc$=896$ proc$\_$cpu$=$tmux proc$\_$mem$=$systemd $\backslash$

temp$=76$ load$=0\mathrm{.}01$

Your log should follow exactly this format $($line breaks in this document for clarity

should not be in your output$)$

Alerts

Your monitor should detect if a parameter matches an alert condition and report this to

both the logfile and stderr. The alert should report the timesamp and the condition broken.

If multiple conditions are broken report them on separate lines. i$.$e:

$20240707-1435$ ALERT: cpu$>80$

Your alerts should follow exactly this format

Bonus $[5$ Points$]$

Make your monitor run indefinitely when the duration is given as $0.$ The command must be

terminated with a ctrl$-$C and print a message that it is stopping to receive this bonus, a

simple infinite loop will not qualify. Hint: man trap

Testing $[20$ Points$]$

Testing system events can be difficult, but not impossible. First write a test plan of the

general proceedure for how you would test monitoring and alerting. $($i$.$e$.$ test cpu usage

by increasing cpu usage and checking for evidence in the log$)$ For now keep this high level

without considering exactly how you will accomplish these tasks. Document as many of

these tests as you think you need, you should try to cover all of the different monitors, but

don$$t have to test only one at a time $($in somecases that might be very difficult or impossible$)$

Call this file testplan$-$initial.txt$.$

Now, for each test use man pages, apropos, and any other resources to develop an idea of

how to create the test condition. For each test case document a command that can create the

conditions you want $($i$.$e$.$ for increasing cpu usage you might run a cpu intensive program$).$

These commands should be usable on the lab machines. Add these to your test plan in a

new file testplan$-$completed.txt

$4$

Lastly execute your test plan, and document the results. Include snippets of the log or

command output which show your monitor passed the test.

$5$

Files to hand in