Append timestamp to log entries generated from AWK

问题:

The script already logs the output to a file called "server_mon.txt. I would like to append a timestamp to each entry for the purpose of tracking server activity.

I now understand that standard AWK doesn't have an inherent time/date function that can easily be assigned to a variable. I attempted the following but didn't work for me:

tail -fn0 /var/log/user | /usr/bin/awk '
BEGIN {
    str = "date +%Y-%m-%d";
    str = | getline date;
    close str;

The following is my full script so far:

#!/bin/bash
PATH=/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin

tail -fn0 /var/log/user | /usr/bin/awk '
    /disconnect_tcp_conn/ { report("down") }
    /daemon apps started/  { report("up") }

    function report(curr_state,    msg) {
        if ( prev_state != curr_state ) {
            msg = "Server is " curr_state
            system("mail -s \047" msg "\047 [e-mail] </dev/null")
            print msg | "cat>&2"
            prev_state = curr_state
        }
    }
'

 &
PID=$!

DIEAT=`expr $SECONDS + 58`
while [ -d /proc/$PID ] && [ "$SECONDS" -lt "$DIEAT" ]
do
        sleep 1
done

[ -d /proc/$PID ] && kill "$PID"
wait

Expect to see a timestamp associated with each log entry to server_mon.txt.

Thanks

回答1:

I highly recommend just reading and printing the timestanps already present in your log file but if that's not an option for some reson then here are you choices:

GUN awk:

$ awk 'BEGIN{ timestamp = strftime("%F %T"); print timestamp }'
2019-05-17 18:40:56

Any awk (much less efficient due to spawning a shell for every call to date):

$ awk 'BEGIN{ cmd="date \"+%F %T\""; timestamp=( (cmd | getline line) > 0 ? line : "N/A"); print timestamp }'
2019-05-17 18:40:59

Put the code where you need to generate the timestamp, I just have it in the BEGIN section to demonstrate how to write the code to generate a timestamp and save it in a variable.

回答2:

Unless I don't understand something about your environment...

systime() returns seconds since Jan 1, 1970

and

mktime(datespec) and strftime([format [, timestamp[, utc-flag]]]) convert that timestamp to useful strings.

See: https://www.tutorialspoint.com/awk/awk_time_functions.htm for details.

回答3:

Assuming each line in the log file is a log entry.

If you want to append timestamp to each line in a file. This is the awk script:

awk '{ print(strftime(), $0); }' input.txt

To learn more about awk time and data funtion read the documentation here.

strftime() function can be configured to your date/time format with parameters.

回答4:

I know there were other answers but in case anyone wants to know what I ended up using and working for my purpose.. here it is:

 msg = "%m/%d/%Y %H:%M:%S 

print strftime(msg) | "tee -a

I ended up using GAWK instead of AWK

This allowed me to have a time and date on the log file in GAWK.

#!/bin/bash
PATH=/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin

tail -fn0 /var/log/user | /usr/bin/gawk '
    /disconnect_tcp_conn/ { report("down") }
    /daemon apps started/  { report("up") }

    function report(curr_state,    msg) {
        if ( prev_state != curr_state ) {
            msg = "%m/%d/%Y %H:%M:%S "Server is " curr_state
            system("mail -s \047" strftime(msg) "\047 [e-mail] </dev/null")
            # print msg | "cat>&2"
            print strftime(msg) | "tee -a \047/var/log/server_mon.txt\047 >&2"
            prev_state = curr_state
        }
    }
'