The safest way to run these queries is by using the backup created by backup-utils loaded into another database server. This database can be quite large and GitHub Enterprise Server can be sensitive to I/O intensive operations that aren't part of anticipated load.
A simple way to do this would be to install a MySQL 5.7 server on the VM receiving the backups and load it automatically. You can then connect to in using root with no password, or whatever you set up for authentication. What this looks like in practice would be similar to this shell script:
# Stop MySQL
sudo systemctl stop mysqld.service
# Unzip the most current backup
gunzip -c /data/current/mysql.sql.gz > /data/mysql.tar
# Untar the current backup
tar xf /data/mysql.tar --directory=/home/github/restore-job/
# Remove the temporary tarball
rm /data/mysql.tar
# Clear the data directory before restoring
sudo rm -rf /var/lib/mysql-data/*
# Run the Percona backup restore
cd /home/github/restore-job && sudo innobackupex --defaults-file=backup-my.cnf --copy-back --datadir=/var/lib/mysql-data .
# Restore the innodb buffer pool
sudo cp -n /var/lib/mysql/ib_buffer_pool /var/lib/mysql-data/
# Restore the innodb data
sudo cp -n /var/lib/mysql/ibdata1 /var/lib/mysql-data/
# Restore the first and second logs
sudo cp -n /var/lib/mysql/ib_logfile0 /var/lib/mysql-data/
sudo cp -n /var/lib/mysql/ib_logfile1 /var/lib/mysql-data/
# Reset ownership
sudo chown -R mysql:mysql /var/lib/mysql-data
# Restore SELinux contexts (if applicable)
sudo restorecon -R /var/lib/mysql-data
# Start MySQL
sudo systemctl start mysqld.service
# Clear the working directory to save some disk space
rm -rf /home/github/restore-job/*