use conky to monitor your vm

flower · 2019-10-17 20:39:06

Hello,

i made a small bash script which monitors my vms. maybe it's useful for you.
vms are expected to run arch and have pacman-contrib installed (this can be changed easily though)
if you place a health.sh script in /root the exit code is evaluated

conkyrc on host:

${execpi 10 ssh root@{HOSTNAME} -C cat /tmp/info}

/etc/systemd/system/sysinfo.service on vm

[Unit]
Description=conkysysinfo

[Service]
ExecStart=/root/info.sh

[Install]
WantedBy=multi-user.target

/root/info.sh on vm

#!/bin/bash

OK="\${color green}"
FAIL="\${color red}"
trap "kill 0" SIGINT

NAME=`cat /etc/hostname`

echo "HEALTH=" > /tmp/info_health
if [ -f /root/health.sh ]; then
	(
		while :
		do
			HEALTH=`/root/health.sh && echo \\\${color green} || echo \\\${color red}`
			echo "HEALTH=\"\\$HEALTH\"" > /tmp/info_health
			sleep 300
		done
	) &
fi

echo "UPDATES=\"0\"" > /tmp/info_updates
(
	while :
	do
		UPDATES=`checkupdates | wc -l`
		echo "UPDATES=\"$UPDATES\"" > /tmp/info_updates
		sleep 3600
	done
) &

MEMTOTAL="0B"
MEMUSED="0B"
SWAPTOTAL=0B""
SWAPUSED="0B"
echo "" > /tmp/info_mem
(
	while :
	do
		printf "MEMTOTAL=\"`free -h | awk '(NR==2){print $2}'`\"\n\
MEMUSED=\"`free -h | awk '(NR==2){print $3}'`\"\n\
SWAPTOTAL=\"`free -h | awk '(NR==3){print $2}'`\"\n\
SWAPUSED=\"`free -h | awk '(NR==3){print $3}'`\"\n" > /tmp/info_mem
		sleep 10
	done
) &

CPUCOUNT="0"
CPUPERCENT="0"
echo "" > /tmp/info_cpu
(
	while :
	do
		echo -e "CPUCOUNT=\"`cat /proc/cpuinfo | grep processor | wc -l`\"\nCPUPERCENT=\"`top -bn10 | awk 'FNR == 3 { print $2+$4; }'`\"\n" > /tmp/info_cpu
		sleep 5
	done
) &

HDDSIZE="0B"
HDDUSED="0B"
echo "" > /tmp/info_hdd
(
        while :
        do
                echo -e "HDDSIZE=\"`df -h | grep vda3 | awk '{print $2}'`\"\nHDDUSED=\"`df -h | grep vda3 | awk '{print $3}'`\"\n" > /tmp/info_hdd
                sleep 300
        done
) &

(
	while :
	do
		(
		. /tmp/info_health
		. /tmp/info_updates
		. /tmp/info_mem
		. /tmp/info_cpu
		. /tmp/info_hdd

		echo -n $HEALTH$NAME\$color
		if [ "$UPDATES" != "0" ]; then
			echo -n "\$alignr$OK$UPDATES Updates\$color "
		fi
		echo
		echo "\$hr"
		echo "RAM \${color grey}$MEMUSED / $MEMTOTAL\$color \${goto 275} SWAP   \${color grey}$SWAPUSED / $SWAPTOTAL \$color"
		echo "CPU \${color grey}$CPUPERCENT% / $CPUCOUNT Threads\$color \${goto 275} HDD    \${color grey}$HDDUSED / ${HDDSIZE} \$color"

		) > /tmp/info
		sleep 1
	done
) &

wait

result:
https://i.imgur.com/1hkScWV.png

moderator edit -- replaced oversized image with link.
Pasting pictures and code

Last edited by 2ManyDogs (2019-10-17 22:15:56)

stanna · 2019-10-17 23:39:08

thanks Flower. this looks like it might come in handy

Trilby · 2019-10-18 01:09:15

You could do a lot to clean up the script on the server - as just one example, rather than calling free and awk several times each while also creating several additional subshells, just call each one once:

#replace these lines
printf "MEMTOTAL=\"`free -h | awk '(NR==2){print $2}'`\"\n\
MEMUSED=\"`free -h | awk '(NR==2){print $3}'`\"\n\
SWAPTOTAL=\"`free -h | awk '(NR==3){print $2}'`\"\n\
SWAPUSED=\"`free -h | awk '(NR==3){print $3}'`\"\n" > /tmp/info_mem
# with this instead
free -h | awk '
   NR==2 { printf "MEMTOTAL=\"%s\"\nMEMUSED=\"%s\"\n", $2, $3; }
   NR==3 { printf "SWAPTOTAL=\"%s\"\nSWAPUSED=\"%s\"\n", $2, $3; }
' >/tmp/info_mem

Also, your loop times are quite odd. Why is the outermost process looping every second to regenerate /tmp/info when all of the content used to create /tmp/info is regenerated at much longer intervals? Specifically, the shortest interval of any component is 5 seconds, yet the /tmp/info is regerated 5 times in that time: once with new data, then 4 times for no reason with stale data.

EDIT: that entire script can be replaced by the following:

Don't bother looping in the script and creating a file that you cat from yet another loop. Just run this from the host:

${execpi 10 ssh root@{HOSTNAME} -C /path/to/this_script}

Then "this_script" on the VM:

#!/bin/sh

# if /tmp/info_updates is absent or older than 60 minutes, (re)create it
[ -z "$(find /tmp -maxdepth 1 -name info_updates -mtime -60)" ] && checkupdates | wc -l >| /tmp/info_updates

cat /etc/hostname
if [ -f /tmp/info_updates ]; then
	read n < /tmp/info_updates
	printf '$alignr${color green}%d Updates$color\n' $n
fi
printf '$hr\n'
free -h | awk '
	NR==2 { printf "RAM ${color grey}%s / %s$color ${goto 275} ", $3, $2; }
	NR==3 { printf "SWAP ${color grey}%s / %s$color\n", $3, $2; }
'
printf 'CPU ${color grey}%d%% / %d Threads$color ${goto 275} ' $(awk '// { print $2 * 100; }' /proc/loadavg) $(nproc)
df -h / | awk 'NR==2 { printf "HDD ${color grey}%s / %s$color\n", $3, $2; }'

Last edited by Trilby (2019-10-18 01:47:39)

qinohe · 2019-10-21 20:56:38

@Trilby, thanks for checkupdates, didn't know about that. Could be I'm missing something obvious but in my simple opinion that script wont work as you expect.
You see, it counts lines after you check for 'mtime'. So in this case you would only see how many updates there are once every 60 min and the line count before checkupdates if you want to see it's output every 10s. Now I'm not into Conky so I have put the logic in2 shell functions with a little add in the form of 'grep' so the output can mean something more, instead of there being a number updates which doesn't say much, you can add a '+' sign behind the number if there is an update you're interested in, see my example: mind the line breaks!
'mtime' should be 'mmin' for min., but I think you meant tha

_upd() {
  [ -f '/tmp/info_updates' ] && awk 'END{print NR}' /tmp/info_updates
  [ -z "$(find /tmp -maxdepth 1 -name info_updates -mmin -60)" ] \
   && checkupdates >| /tmp/info_updates
}

# if an update of interest, put '+' sign after number of updates(former function)
_ifupd() {
  [ -f '/tmp/info_updates' ] && grep -Ew 'apache|bash|linux|mariadb|php|systemd' \
   /tmp/info_updates && echo +
}

Tested on tmux statusbar

edit: add check

Last edited by qinohe (2019-10-21 21:09:38)

Trilby · 2019-10-21 22:33:56

qinohe wrote:

it counts lines after you check for 'mtime'. So in this case you would only see how many updates there are once every 60 min and the line count before checkupdates if you want to see it's output every 10s.

Exactly, that's the point. Checkupdates doesn't need to be run every 10 seconds. It runs every 60 minutes, and the number of available updates are stored in the temp file.

There is no reason to store the list to the temp file every 60 minutes and count the number of lines every 10 seconds; that count couldn't possibly change. Instead get the list and count the lines every 60 minutes, and just read that number every 10 seconds.

flower · 2019-10-21 23:32:11

Thanks for your input @Trilby
I'll definitly learned something about bash today and will integrate your ideas

qinohe · 2019-10-22 00:53:11

Trilby wrote:

..., and just read that number every 10 seconds.

Ah yeah thanks, see, I'm missing something obvious:(
anyway @flower if you wish you could still use that second function(_ifupd) if you have use for it;)

Arch Linux

#1 2019-10-17 20:39:06

use conky to monitor your vm

#2 2019-10-17 23:39:08

Re: use conky to monitor your vm

#3 2019-10-18 01:09:15

Re: use conky to monitor your vm

#4 2019-10-21 20:56:38

Re: use conky to monitor your vm

#5 2019-10-21 22:33:56

Re: use conky to monitor your vm

#6 2019-10-21 23:32:11

Re: use conky to monitor your vm

#7 2019-10-22 00:53:11

Re: use conky to monitor your vm

Board footer