Embedded PostgreSQL and Redis

Introduction

To deploy the GitGuardian app, a PostgreSQL instance and a Redis instance are required. This page is dedicated to helping you set up and managed an embedded one.

Embedded databases are great for testing purposes and small-scale infrastructures, but we don't recommend them for a more robust setup.

High-Availability

Embedded databases are not highly available.

Installation

Select "Embedded Postgres/Redis" on the configuration page during the installation, and choose the disk size if needed. But in most cases, the default disk size is enough.

Upsizing the disk

On an existing cluster

If you installed the GitGuardian app on one of your clusters, you need to follow these steps:

Make sure your default storage class handles volume expansion
Check the allowVolumeExpansion flag for your default storage class. To get information about your storage classes, run kubectl get storageclasses:

> kubectl get storageclasses.storage.k8s.io
NAME                PROVISIONER          RECLAIMPOLICY   VOLUMEBINDINGMODE   ALLOWVOLUMEEXPANSION   AGE
default (default)   ceph.rook.io/block   Delete          Immediate           false                  2d17h

If needed, run kubectl patch storageclass <your-storage-class-name> -p '{"allowVolumeExpansion": true}'.
Go to the KOTS Admin Console, select the "Config" tab, edit the disk size, save the new configuration, and deploy it.
Upsizing will be done on the next restart of the PostgreSQL or Redis container. You can force it by deleting the pod. This will cause a small interruption in the application.

On an embedded cluster

On an embedded cluster, you must first check the installed version of rook-ceph. To get it, run kubectl -n rook-ceph exec $(kubectl -n rook-ceph get pod -l app=rook-discover -o name | head -n1) -- rook version.

Rook version >= 1.4.3

Follow these steps:

Set the allowVolumeExpansion flag to true with kubectl patch storageclass default -p '{"allowVolumeExpansion": true}'
Go to the KOTS Admin Console, select the "Config" tab, edit the disk size, save the new configuration, and deploy it.
Upsizing will be done on the next restart of the PostgreSQL or Redis container. You can force it by deleting the pod. This will cause a small interruption in the application.

Rook version 1.0.4

This version of Rook Ceph does not handle volume expansion through Kubernetes. It's also not easily upgradeable. You can follow the procedure below to increase the size of your PostgreSQL or Redis disk. If you wish to upgrade from rook 1.0.4 to rook >= 1.4.3, you will need to do a backup/restore of your instance. To do this, follow this procedure.

Before continuing, please make sure you have functional backups. Documentation is available here

First, make sure the package e2fsprogs is installed on your instance.

To upsize the disk for PostgreSQL, run these commands. You can change the disk size (50G here), by using the same value in the first two commands.

kubectl -n rook-ceph exec $(kubectl -n rook-ceph get pod -l app=rook-ceph-operator -o name | head -n1) -- rbd -p replicapool resize $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep postgresql-data | awk '{print $2}') --size 50G
kubectl -n rook-ceph exec $(kubectl -n rook-ceph get pod -l app=rook-ceph-operator -o name | head -n1) -- rbd -p replicapool resize $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep postgresql-backup | awk '{print $2}') --size 50G

sudo resize2fs /dev/$(lsblk | grep $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep postgresql-data | awk '{print $2}') | awk '{print $1}')
sudo resize2fs /dev/$(lsblk | grep $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep postgresql-backup | awk '{print $2}') | awk '{print $1}')

To upsize the disk for Redis, run these commands. You can change the disk size (20G here).

kubectl -n rook-ceph exec $(kubectl -n rook-ceph get pod -l app=rook-ceph-operator -o name | head -n1) -- rbd -p replicapool resize $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep redis-data | awk '{print $2}') --size 20G

sudo resize2fs /dev/$(lsblk | grep $(kubectl -n default get pvc -o custom-columns=NAME:.metadata.name,VOLUME:.spec.volumeName | grep redis-data | awk '{print $2}') | awk '{print $1}')