PgBouncer K8s
- Canonical
- Databases
Channel | Revision | Published | Runs on |
---|---|---|---|
1/stable | 269 | 11 Sep 2024 | |
1/stable | 268 | 11 Sep 2024 | |
1/candidate | 269 | 09 Sep 2024 | |
1/candidate | 268 | 09 Sep 2024 | |
1/beta | 269 | 09 Sep 2024 | |
1/beta | 268 | 09 Sep 2024 | |
1/edge | 342 | Yesterday | |
1/edge | 341 | Yesterday |
juju deploy pgbouncer-k8s --channel 1/stable
Deploy Kubernetes operators easily with Juju, the Universal Operator Lifecycle Manager. Need a Kubernetes cluster? Install MicroK8s to create a full CNCF-certified Kubernetes system in under 60 seconds.
Platform:
Scale your PgBouncer K8s
This is part of the PgBouncer K8s Tutorial. Please refer to this page for more information and the overview of the content.
Adding and Removing units
Please check the explanation of scaling Charmed PostgreSQL K8s operator here.
Add more PgBouncer instances
You can add two more units to your deployed PgBouncer application by scaling it using:
juju scale-application pgbouncer-k8s 3
You can now watch the scaling process in live using: juju status --watch 1s
. It usually takes several minutes for new cluster members to be added. You’ll know that all three nodes are in sync when juju status
reports Workload=active
and Agent=idle
:
Model Controller Cloud/Region Version SLA Timestamp
test16 microk8s microk8s/localhost 3.1.6 unsupported 22:16:20+02:00
App Version Status Scale Charm Channel Rev Address Exposed Message
data-integrator active 1 data-integrator stable 13 10.152.183.136 no
pgbouncer-k8s 1.18.0 active 3 pgbouncer-k8s 1/stable 76 10.152.183.84 no
postgresql-k8s 14.9 active 1 postgresql-k8s 14/stable 158 10.152.183.92 no Primary
Unit Workload Agent Address Ports Message
data-integrator/0* active idle 10.1.12.16
pgbouncer-k8s/0* active idle 10.1.12.15
pgbouncer-k8s/1 active idle 10.1.12.61
pgbouncer-k8s/2 active idle 10.1.12.50
postgresql-k8s/0* active idle 10.1.12.6 Primary
You can scale Charmed PostgreSQL in the same way:
juju scale-application postgresql-k8s 3
Make sure all units are active using juju status
:
Model Controller Cloud/Region Version SLA Timestamp
test16 microk8s microk8s/localhost 3.1.6 unsupported 22:18:00+02:00
App Version Status Scale Charm Channel Rev Address Exposed Message
data-integrator active 1 data-integrator stable 13 10.152.183.136 no
pgbouncer-k8s 1.18.0 active 3 pgbouncer-k8s 1/stable 76 10.152.183.84 no
postgresql-k8s 14.9 active 3 postgresql-k8s 14/stable 158 10.152.183.92 no
Unit Workload Agent Address Ports Message
data-integrator/0* active idle 10.1.12.16
pgbouncer-k8s/0* active idle 10.1.12.15
pgbouncer-k8s/1 active idle 10.1.12.61
pgbouncer-k8s/2 active idle 10.1.12.50
postgresql-k8s/0* active idle 10.1.12.6 Primary
postgresql-k8s/1 active idle 10.1.12.35
postgresql-k8s/2 active idle 10.1.12.39
Remove extra members
Removing a unit from the application scales the replicas down.
juju scale-application pgbouncer-k8s 2
juju scale-application postgresql-k8s 2
You’ll know that the replica was successfully removed when juju status --watch 1s
reports:
Model Controller Cloud/Region Version SLA Timestamp
test16 microk8s microk8s/localhost 3.1.6 unsupported 22:19:10+02:00
App Version Status Scale Charm Channel Rev Address Exposed Message
data-integrator active 1 data-integrator stable 13 10.152.183.136 no
pgbouncer-k8s 1.18.0 active 2 pgbouncer-k8s 1/stable 76 10.152.183.84 no
postgresql-k8s 14.9 active 2 postgresql-k8s 14/stable 158 10.152.183.92 no
Unit Workload Agent Address Ports Message
data-integrator/0* active idle 10.1.12.16
pgbouncer-k8s/0* active idle 10.1.12.15
pgbouncer-k8s/1 active idle 10.1.12.61
postgresql-k8s/0* active idle 10.1.12.6 Primary
postgresql-k8s/1 active idle 10.1.12.35