TODO.md

Immediate

pg-playbook

1. Create image as part of the pg-playbook
2. Tuning for Pg / OS
   • max_connections
   • shared_buffer
   • checkpoint_timeout
3. No pg logs in the directory anymore - is it going to syslog?
4. Verify the postgresql.base.conf defaults

patroni setup

2. Why are stderr from patroni starting up postgres not visible in the patroni logs
3. Necessary setup for

• bootstrapping a brand new cluster
  • tweak the bootstrap config - needs to create relevant users (replicator, postgres user with password?)
• bootstrapping an existing Pg instance as the master

4. patroni.service file

• Dynamically update the IP based on a ExecStartPre ?
  • https://superuser.com/questions/968561/how-to-get-the-machine-ip-address-in-a-systemd-service-file

ansible setup

1. Get ansible to use dynamic inventory
2. Get host_key_checking to work
3. Group vars names using illegal char '-'

Longer term

1. Further tuning for Pg instnace

• os optimizations
• network optimizations
• pd-ssh / fs optimizations

2. Logging

• get patroni, etcd and postgresql logs streaming to cloudlogging
• Using stackdriver logging for logs

3. Monitoring

   • 3rd party tool?
   • what should be monitored - rep lag, disk, io

4. backups

• wall-g, pg_backrest
• simplify the bootstrap

5. connection pooling

• pgbouncer/pg-pool?
• run on dedicated IG
• allow the use of only 1 ILB

6. security

• ssl connections
• user lockdown

Steps to handle the following operational tasks:

1. Upgrading disk size
2. Updating os disk (same major version)
3. Updating new major version