-
Notifications
You must be signed in to change notification settings - Fork 1k
Closed
Description
We recently set up a postgres cluster on AWS on a kubernetes cluster set up with Typhoon using this configuration:
kind: "postgresql"
apiVersion: "acid.zalan.do/v1"
metadata:
name: "iot-historics"
namespace: "postgres-operator"
labels:
team: iot
spec:
teamId: "iot"
postgresql:
version: "12"
numberOfInstances: 2
enableMasterLoadBalancer: true
enableReplicaLoadBalancer: true
volume:
size: "200Gi"
users:
iot-historics: []
preparedDatabases:
iot_historics:
defaultUsers: true
extensions:
timescaledb: public
resources:
requests:
cpu: 1000m
memory: 2048Mi
limits:
cpu: 1000m
memory: 2048Mi
Operator version is 1.5.0
installed with helm.
Our operator has an IAM role attached, granting full ec2
and ebs
permissions (this is of course too broad, but we had permission issues during this test).
Since we hit the limit of the disk, we tried to expand the volume by changing the size in the manifest and reapplying it. Unfortunately this seems to fail. The operator logs show
time="2020-07-08T07:56:37Z" level=debug msg="syncing persistent volumes" cluster-name=postgres-operator/iot-historics pkg=cluster
time="2020-07-08T07:56:38Z" level=debug msg="updating persistent volume \"pvc-bae2f27e-ac44-42ba-a5f7-aa24aaa8becc\" to 300" cluster-name=postgres-operator/iot-historics pkg=cluster
time="2020-07-08T07:56:39Z" level=warning msg="error while syncing cluster state: could not sync persistent volumes: could not sync volumes: could not resize EBS volume \"vol-0b0407e7caa5e083a\": could not get information about the volume: InvalidVolume.NotFound: The volume 'vol-0b0407e7caa5e083a' does not exist.\n\tstatus code: 400, request id: 6804d969-3971-46a6-b0d3-66cda100a629" cluster-name=postgres-operator/iot-historics pkg=cluster
When we check the AWS console, the volume does exist.
When we update the PVC manually, the resize gets executed succesfully though.
Any idea what might be causing this?
Metadata
Metadata
Assignees
Labels
No labels