Here are some scripts we wrote at squidsolutions.com; they help us administrate the AWS Redshift clusters we are in charge of.
The bin/redshift.sh script ensures that a cluster's configuration and data are saved upon restart. In fact, stopping a redshift cluster is more like a cluster deletion. If you don't do it properly, you will loose:
- The cluster's data
- The cluster's network configuration (In a VPC ? In which subnet group ? With which security group ? etc.)
- The cluster's parameter group name
bin/redshift.sh automatically creates a final data snapshot and saves these informations in a JSON file. During the cluster start, the script will restore these informations. Run from a crontab, it can help you save a lot of money.
You must have awscli and jq installed:
pip install awscli
apt-get install jq
The etc/redshift.conf file handles a few configuration variables. See the inline comments for more details.
./bin/redshift.sh
Usage: ./bin/redshift.sh -a {start|stop|save} -r aws_region -c cluster_name [-p awscli-profile]
- -a: The action to take,
- -r: The AWS region in which the cluster is defined,
- -c: The name of the cluster,
- -p: The name of the profile used by awscli.
The stop and save actions will create a .json file in the etc/ directory for each cluster you need to start/stop.
Say you want your dev cluster up and running at working hour (not during the week-end):
0 20 * * 1-5 /opt/squid/aws-redshift-shell/bin/redshift.sh -a stop -r us-east-1 -c dev-cluster -p myCompany
0 8 * * 1-5 /opt/squid/aws-redshift-shell/bin/redshift.sh -a start -r us-east-1 -c dev-cluster -p myCompany
Note that, depending on the cluster's size, it may take time to be fully started.
The shell scripts inside the bin/ directory are licensed under the GNU General Public License (GPL) v.3.
- Grégoire Doumergue https://github.com/gdoumergue