Correctly grep and display the uptime, load average and amount of users?

Raymii

Because of the new version of my status monitoring script I've been busy rewriting the client script a little bit. See here

Now I find it quite hard to reliably parse the output of the uptime command with only bash/sed/awk etc. I have a solution which works on my tests (iMac, Debian 6, RHEL 5). Maybe peoples can test it to see if they also get the correct output?

https://raymii.org/cms/p_Get_uptime_load_and_users_with_grep_sed_and_awk

bnmkl

@Raymii:

I prefer what you are doing (parsing uptime), but if you are having problems making it "cross-distro", you can try to get the info directly. I wrote these line on my machine only, but in theory they should work on all.

USERS: ( 2 users )
echo "$(who | cut -d ' ' -f1 | sort | uniq | wc -l) users"

UPTIME: ( 5 days, 19:12:41 )
upsecs=$(cat /proc/uptime | cut -d ' ' -f1); echo "$((`echo $upsecs | cut -d '.' -f1` (60*60*24))) days, $(date -d "`echo @$upsecs`" +%H:%M:%S)"

LOAD AVERAGE: ( 0.03 0.02 0.00 )
for avload in $(cat /proc/loadavg); do if [[ $avload =~ ^.*\\..*$ ]]; then echo -n  "$avload "; fi done

You should be able to make my lines better with your awk skills.

corehosting

Only with awk ones tested on Linux and MAC
uptime: uptime | awk '{sub(/\,/,"");print $3" "$4 }'
users: uptime | awk '{gsub(/\,/,"");print $6" "$7 }'
load: uptime | awk '{gsub(/\,/,"");print $10 }'

Awk by default doesn't care too much of the number of spaces separating entries

Raymii

@bnmkl the /proc filesystem is not available on every system I'm using.

@corehosting the amount of variables differs if a host is up less then a day.

corehosting

@Raymii said: @corehosting the amount of variables differs if a host is up less then a day.

True... Don't have anything with less than a day to test that out

Raymii

@corehosting said: True... Don't have anything with less than a day to test that out

Reboot like a sir! Also that is mostly fixed with an extra [if...fi]...

William

@corehosting said: True... Don't have anything with less than a day to test that out

root@sg:~# uptime | awk '{sub(/\,/,"");print $3" "$4 }'
2:20 1

Raymii

@William what do the commands I use give?:

uptime | grep -ohe 'up .*' | sed 's/,//g' | awk '{ print $2" "$3 }'

uptime | grep -ohe '[0-9.*] user[s,]'

uptime | grep -ohe 'load average[s:][: ].*' | awk '{ print $3 }'

And, what OS?

corehosting

[root@curse ~]# uptime | awk '{sub(/\,/,"");print $3" "$4 }'
1:06 1
[root@curse ~]# uptime | grep -ohe 'up .' | sed 's/,//g' | awk '{ print $2" "$3 }'
1:06 1
[root@curse ~]# uptime | grep -ohe '[0-9.] user[s,]'
1 user,
[root@curse ~]# uptime | grep -ohe 'load average[s:][: ].*' | awk '{ print $3 }'
0.00,

EL6.3

Raymii

@Corehosting thank you. CentOS/SL or real RHEL? My test on RHEL 6.3 gives me funky output after a reboot. Hope it is different tomorrow.

corehosting

@Raymii: SL to be precise, but the result should be the same
Which version of procps ? So that we can compare.

bnmkl

@Raymii... "The /proc filesystem is not available on every system I'm using."

Does that mean you are using some systems that have an alternate uptime command @Raymii? If so, then you must already have to test which system you are on to determine which command to use. Thus, I would still use /proc to cover "cross-distro" for the ones that do as a function, then however many (least) other functions to cover the "system-sphere".

The last commands you gave did not work correctly. My uptime format is:

uptime
 23:06:54 up 6 days,  2:34, load average: 0.00, 0.00, 0.00

It does seem proven from your own words really that it can not be achieved by just passing uptime as it changes too much. Though I like your determination to keep trying! It is nice to have precise code, and sometimes, the code is more admirable than the application.