Using awk output to run mysqldump - linux

Question

I am trying to learn more about shell scripts and have pieced this together if someone would please educate/critique the use of awk in particular. Because it's based on each value of the executed query, is this a viable option? Is there a better method, ie. for or while loop?
timestamp = $(date "+%Y=%m-%d %H:%M:%S")
user_dir = "server_dir_name"
backup_path = "/home1/$user_dir/public_html"
local_bckup_store = "/home1/$user_dir/backups"
db_prefix = "database_prefix"
db_user = "database_username"
db_pwd = "database_user_password" //used here for simplicity, not security
tar -zcvf "$local_bckup_store/$user_dir-public_html-$timestamp.tar.gz" $backup_path
mysql -NBr -u$db_user -p$db_pwd -Dinformation_schema -e "SELECT schema_name FROM schemata WHERE schema_name LIKE '/$db_prefix_%/"';"
| awk '{ system("mysqldump -u$db_user -p$db_pwd $1 > $local_bckup_store/$1.$timestamp.sql | gzip" $1) }'
I will address the security issue of storing the *$db_pwd*, even though this will be stored and run locally on the server, but would appreciate some input on best practices.
Thanks.

Answer 1

You're really just using awk to obtain the first parameter returned from your mysql statement.
I would recommend saving that param in a variable and then doing what you want with it.
UPDATED Example (I think this is what you want):
e.g.
while read schema_name ; do
backup_file=${local_bckup_store}/${schema_name}.${timestamp}.sql
mysqldump -u${db_user} -p${db_pwd} ${schema_name} > ${backup_file}
gzip ${backup_file}
done < `mysql -NBr -u${db_user} -p${db_pwd} -Dinformation_schema -e "SELECT schema_name FROM schemata WHERE schema_name LIKE 'db_prefix_%';" | awk '{ print $1 }'`
Note the backticks, that runs the command and returns the STDOUT text, which will feed into the STDIN of the 'while read' line. Also, I've put the shell variable names in parenthesis, this is also a good practice to help keep them from being incorrectly interpolated.
This way, you avoid awk having to exec another shell just to run the mysqldump command and your code is [slightly] easier to maintain. As an added bonus, you can add error checking to be sure the mysql command succeeded before calling mysqldump.
Additionally, I would recommend using 'cut' instead of 'awk' for this, it's a bit more efficient for what you're trying to do.
FYI, you could also send the STDOUT of your mysqldump directly into the STDIN of gzip like such:
mysqldump -u${db_user} -p${db_pwd} ${schema_name} | gzip > ${backup_file}.gz

Answer 2

This awk command is problematic:
awk '{ system("mysqldump -u$db_user -p$db_pwd $1 > $local_bckup_store/$1.$timestamp.sql | gzip" $1) }'
Since awk cannot use shell variables $db_user etc like that in single quoted. To pass shell variables to awk use this syntax:
awk -v $db_user="$db_user" -vdb_pwd="$db_pwd" ts="$timestamp" '{
system("mysqldump -udb_user -pdb_pwd " $1 " > $local_bckup_store/" $1 "." ts ".sql | gzip" $1) }'
PS: You mysqldump | awk is stil untested here.
Another option is to do it directly in shell as:
schema=$(mysql -NBr -u$db_user -p$db_pwd -Dinformation_schema -e "SELECT schema_name FROM schemata WHERE schema_name LIKE 'db_prefix_%';")
mysqldump -u$db_user -p$db_pwd "$schema" > "$local_bckup_store/$schema.$timestamp.sql"

Answer 3

It's a good idea, but you really arent doing enough text parsing you warrant awk, you can do it all in bash like this.
query = "SELECT schema_name FROM schemata WHERE schema_name LIKE 'db_prefix_%';"
mysql -NBr -u$db_user -p$db_pwd -Dinformation_schema -e $query | while read line
do
arr=($line)
mysqldump -u$db_user -p$db_pwd ${arr[0]} > $local_bckup_store/${arr[0]}.$timestamp.sql && gzip $local_bckup_store/${arr[0]}.$timestamp.sql
done
bash will create an array by splitting the string on whitespace for you (this is the point of arr=($line)).