0

I am backing small databases from several websites and sending the admin an email with the results. I have already directed stderr to send the complete error in the email. Is there a way I can detect whether an error occurred so that I can change the email subject to match?

I am using two files, the first just sets things up and is different in each site.

#!/usr/bin/bash
#set -euo pipefail
set -u

email="admin@example.com"
keep="+14"


#include script
. cronjobs/do_backup.sh


echo Main
#mysql Config File
dbcfg="cronjobs/_dbaccess.cnf"


#Backup using database, group_suffix in cnf file and backup_name
db="zzzcity_directory"
dbgrp="_zzz_directory"
bkp="db_zzz_directory.sql"
backup $db $dbcfg $dbgrp $bkp &>>$msg


#Backup using database, group_suffix in cnf file and backup_name
db="zzzcity_potluck"
dbgrp="_zzz_potluck"
bkp="db_zzz_potluck.sql"
backup $db $dbcfg $dbgrp $bkp &>>$msg


#Maintain backup copies and send email
finalize $keep $email
echo Done

The second one is identical on every site.

do_backup.sh

#!/usr/bin/bash

timestamp=$(date +"%F")
backup_dir="backup"
mysqldump="/usr/bin/mysqldump"
stat="/usr/bin/stat"
msg="tmp/backup_msg"


echo "Date:     $(date)" > $msg
echo "Hostname: $(hostname)" >> $msg
echo >> $msg
echo >> $msg
mkdir -p "$backup_dir/$timestamp" 2>> $msg


main() {
  db=$1
  dbcfg=$2
  dbpgrp=$3
  bkp="$backup_dir/$timestamp/$4"
  echo "Backing:  $db"
  echo "To:       $bkp"
  $mysqldump --defaults-extra-file=$dbcfg --defaults-group-suffix=$dbgrp $db > $bkp
  size=$($stat -c%s $bkp)
  echo "Filesize: $size"
  echo
}


backup () {
  main $1 $2 $3 $4
}


finish () {
  keep=$1
  echo "Keeping:  $keep copies"
  echo "Removing excess backups"
  find $backup_dir/ -type d -mtime +$keep -exec rm -r {} +
  echo
  echo "Finished: $(date)"
}


finalize (){
  finish $1 &>>$msg
  mail  -s "MySQL Backup script has run" "$2" <$msg
  #rm -f $MSG
}

Further explanation

  • It's run using cron, once a day at 2 am
  • It's backing up MySQL databases with authentication in a .cnf file
  • I am keeping 14 copies in directories whose names are the date of backup
  • The oldest gets deleted

The email looks like this

Date: Tue May 9 02:00:02 AEST 2023
Hostname: zzz

Backing: zzzcity_directory
To: backup/2023-05-09/db_zzz_directory.sql
Filesize: 6463

Backing: zzzcity_potluck
To: backup/2023-05-09/db_zzz_potluck.sql
Filesize: 3541

Keeping: +14 copies
Removing excess backups

Finished: Tue May 9 02:00:04 AEST 2023

Question
At the moment, the subject of the email is static. I would like to change it if there is an error. Otherwise, I have to open all the emails. Since I have already used stderr to build up the details in the email, how do I obtain an indication that something has failed? Two that come to mind are mysqldump or filesystem errors.

Rohit Gupta
  • 375
  • 2
    What do you want to check for? If mysqldump ran correctly? Take a look at the return code, should be 0 if everything worked. – Panki May 09 '23 at 07:48
  • 1
    Please [edit] and clarify what the actual question is. Your title mentions detecting an error, so I guess you want to detect some sort of error? What error? What should happen? What does using stderr have to do with it? – terdon May 09 '23 at 11:59
  • @terdon - done. I am not an expert at linux. I thought stderr contains any errors in the commands. in this case from mysqldump and from backup script, which I have redirected to a file. – Rohit Gupta May 10 '23 at 13:14
  • Side note: quote right, unless you have a very, very good reason not to quote. Hardly any reason is good enough though. – Kamil Maciorowski May 10 '23 at 14:20

0 Answers0