Bash : How to tail latest file in continuous loop

842 Views Asked by At

I am trying to write script that monitors log file and on specific error it performs certain function. I got the main code working however stuck with one problem. Application create new log file when system date changes (log file name pattern is LOG.NODE1.DDMMYYYY), how do I get my code to automatically switch to new file that is created. Following is my script so far,

#!/bin/sh

logfile=$(ls -t $DIR"/env/log/LOG"* | head -n 1)
echo $logfile
tail -f $logfile | while read LOGLINE
do

if [[  "${LOGLINE}" == *";A database exception has occurred: FATAL DBERR:  SQL_ERROR: ORA-00001: unique constraint (IX_TEST1) violated"* ]];

then
        #Do something

fi
done
1

There are 1 best solutions below

1
Charles Duffy On BEST ANSWER
#!/bin/bash
#      ^^^^ -- **NOT** /bin/sh

substring=";A database exception has occurred: FATAL DBERR:  SQL_ERROR: ORA-00001: unique constraint (IX_TEST1) violated"
newest=
timeout=10  # number of seconds of no input after which to look for a newer file

# sets a global shell variable called "newest" when run
# to keep overhead down, this avoids invoking any external commands
find_newest() {
  set -- "${DIR?The variable DIR is required}"/env/log/LOG*
  [[ -e $1 || -L $1 ]] || return 1
  newest=$1; shift
  while (( $# )); do
    [[ $1 -nt $newest ]] && newest=$1
    shift
  done
}

while :; do
  find_newest  # check for newer files
  # if the newest file isn't the one we're already following...
  if [[ $tailing_from_file != "$newest" ]]; then
    exec < <(tail -f -- "$newest")  # start a new copy of tail following the newer one
    tailing_from_file=$newest       # and record that file's name
  fi
  if read -t "$timeout" -r line && [[ $line = *"$substring"* ]]; then
    echo "Do something here"
  fi
done