Read logs real time with bash

I have this in the system logs:

Feb 27 02:53:51 Latitude-E6430 rsyslogd: action up   7
Feb 27 02:53:51 Latitude-E6430 rsyslogd: action down B
Feb 27 02:53:51 Latitude-E6430 rsyslogd: action up   U
....etc

I want to make a script that is monitoring the logs full time as a service, but I don't know how to do the condition.

1 - That when you read, line by line, store it in a variable.

2 - read variable if it finds in the line, the words [rsyslogd] and [7] it shows [echo "Found" ]

It would be a loop to read the file line by line, looking for that pattern of words.

Example Script

path_full_log="/home/full.log"


function reader_time_real(){

    while read full_log; do
        chek7=$(cat $full_log | grep "rsyslogdd" | grep 7 )
            if [[ $? = 0 ]]; then 
                echo "Found 7";

            fi

        chekB=$(cat $full_log | grep "rsyslogdd" | grep B )
            if [[ $? = 0 ]]; then 
                echo "Found B";

            fi
    done < $path_full_log
}

Your script is not suitable to do what you want to do. It reads the file only once, but completely. If you loop this, you won't get only new output and it would be a performance nightmare.

Also, you have a useless use of cat, you grep a fixed string with a regex pattern and save the result in a variable you don't use anymore. Also you should not use unquoted filename variables.

Use tail -f | grep --line-buffered:

tail -fn0 "$path_full_log" \
| grep -F --line-buffered 'rsyslogdd' \
| grep -Eo --line-buffered '(7|B)' \
| while read line; do
      echo "Found $line"
  done

Or using xargs:

tail -fn0 "$path_full_log" \
| grep -F --line-buffered 'rsyslogdd' \
| grep -Eo --line-buffered '(7|B)' \
| xargs -I{} echo "Found {}"

• tail -fn0 reads new lines of the file when they are added.
• grep -F searches for a fixed string in that line.
• --line-buffered tells grep to read lines when they come in instead of wait for the end of the input stream which is default.
• grep -Eo searches for the ERE pattern (-E) (7|B) and outputs only the matched string (-o).
• while read line or xargs -I{} takes that output and runs your command on the output (here: echo).

I've made the following script, which updates every second, and highlights any new log entries. It also defines the area in which the log should be displayed (with the logrows and logcols variables), so it's made for use in a fixed size terminal window (I'm using it inside tmux).

You could probably modify this script to highlight only lines that fit a specific criteria, as you indicate.

#!/bin/bash

# Initial configuration variables
logrows=31
logcols=127

# Local function definition
printnew() {
  clear
  echo -e "\e[0;30;47m"
  hightail=$(tail -n "$logrows" /var/log/syslog | cut -b "1-$logcols")
  echo -ne "${hightail}\e[0m"
}

printdif() {
  (( newrows = logrows - logdiff ))
  clear
  normtail=$(tail -n "$logrows" /var/log/syslog | head -n "$newrows" | cut -b "1-$logcols")
  echo -n "$normtail"
  echo -e "\e[0;30;47m"
  hightail=$(tail -n "$logdiff" /var/log/syslog | cut -b "1-$logcols")
  echo -ne "${hightail}\e[0m"
}

printold() {
  clear
  normtail=$(tail -n "$logrows" /var/log/syslog | cut -b "1-$logcols")
  echo -n "$normtail"
}

# Loop every second
while true
do
  sleep 1 &

  # Read log sizes
  logsize=$(wc -l /var/log/syslog | cut -d ' ' -f1)
  (( logdiff = logsize - presize ))

  # If more than $logrows lines are new
  if (( logdiff > "$logrows" ))
  then
    printnew
    redraw=1
  # If less than $logrows lines but more than 0 lines are new
  elif (( logdiff > 0 ))
  then
    printdif
    redraw=1
  # If no lines are new
  else
    if [[ "$redraw" -eq 1 ]]
    then
      printold
      redraw=0
    fi
  fi

  presize="$logsize"
  wait # for sleep
done

The script is made with 3 scenarios in mind:

1. The entire contents of logrows is new - everything is highlighted (printnew).
2. Some lines of the log is new - these are highlighted (printdif).
3. No new log lines - nothing is highlighted (printold).

Disclaimer: The code could probably be optimized, but I'm generally more concerned with my code being readable rather than as condensed as possible.