During restore tests
...
recovery:
source: restore
database: app
secret:
name: restore-app
# recoveryTarget:
# targetTime: "2025-05-16 10:21:25.723492+02"
externalClusters:
- name: restore
plugin:
name: barman-cloud.cloudnative-pg.io
parameters:
barmanObjectName: restore-c800
serverName: restore
...
I disabled the plugin by setting enabled and isWALArchiver to false
...
plugins:
- name: barman-cloud.cloudnative-pg.io
enabled: false
isWALArchiver: false
parameters:
barmanObjectName: restore-c800
...
apiVersion: barmancloud.cnpg.io/v1
kind: ObjectStore
metadata:
name: restore-c800
spec:
retentionPolicy: "7d"
configuration:
destinationPath: "s3://xxx"
endpointURL: "https://sxxx"
endpointCA:
name: xxx-ca
key: tls.crt
s3Credentials:
accessKeyId:
name: xxx-creds
key: ACCESS_KEY_ID
secretAccessKey:
name: xxx-creds
key: ACCESS_SECRET_KEY
wal:
compression: gzip # bzip2;gzip;snappy
maxParallel: 2
data:
compression: gzip # bzip2;gzip;snappy
immediateCheckpoint: true
jobs: 2
but the plugin still complains about
ERROR: WAL archive check failed for server restore: Expected empty archive
As a workaround, I commented out the plugin
...
# plugins:
# - name: barman-cloud.cloudnative-pg.io
# enabled: false
# isWALArchiver: false
# parameters:
# barmanObjectName: restore-c800
...
After that, the restore is working.
Then comment it in (not enabled)
plugins:
- name: barman-cloud.cloudnative-pg.io
enabled: false
isWALArchiver: false
parameters:
barmanObjectName: restore-c800
no error
enabling it
plugins:
- name: barman-cloud.cloudnative-pg.io
enabled: true
isWALArchiver: true
parameters:
barmanObjectName: restore-c800
no error
select pg_switch_wal();
no error, the not empty archive is being used.
IMO if the plugin is disabled and not used as WALArchiver, it should not check for an empty archive but during enabling it should.
But I the plugin would behave like this, what should be the consequence during enabling on an not empty archive?
Shutting it down the cluster is maybe also not an elegant way...
What do you think?
During restore tests
I disabled the plugin by setting enabled and isWALArchiver to false
but the plugin still complains about
ERROR: WAL archive check failed for server restore: Expected empty archiveAs a workaround, I commented out the plugin
After that, the restore is working.
Then comment it in (not enabled)
no error
enabling it
no error
select pg_switch_wal();no error, the not empty archive is being used.
IMO if the plugin is disabled and not used as WALArchiver, it should not check for an empty archive but during enabling it should.
But I the plugin would behave like this, what should be the consequence during enabling on an not empty archive?
Shutting it down the cluster is maybe also not an elegant way...
What do you think?