Percona-xtradb-cluster-operator 1.8.0 cluster restore from s3 does not download all folders

Hello

We have a healtly cluster that has backups enabled on s3 storage.
We use https://dcd.ionos.com/
Backups are being processed hourly and every backup takes no more than 5 min.

Problem is with restore.

After creating CRD/PerconaXtraDBClusterRestore proces is initiated fine.
But when downloading files from s3 there are missing folders. But no errors from log.
The backup folder on s3 contains all necessary folders and files.

Folders on s3 storage:

[2021-10-14 15:25:06 EEST]     0B common/
[2021-10-14 15:25:06 EEST]     0B customer@002d1/
[2021-10-14 15:25:06 EEST]     0B customer@002d2/
[2021-10-14 15:25:06 EEST]     0B customer@002d3/
[2021-10-14 15:25:06 EEST]     0B mysql/
[2021-10-14 15:25:06 EEST]     0B performance_schema/
[2021-10-14 15:25:06 EEST]     0B sys/. <-- This was NOT copied during restore
[2021-10-14 15:25:06 EEST]     0B testabaze/   <-- This was NOT copied during restore

From LOG

+ xbcloud get s3://customer-db-px-2021-10-13-12:05:12-full --parallel=10
++ grep -c processor /proc/cpuinfo
+ xbstream --decompress -x -C /datadir/pxc_sst_rHi2 --parallel=8
211013 12:59:58 xbcloud: Successfully connected.
211013 12:59:58 xbcloud: Downloading
>> All files are downloaded and successfull notice is printed.
>> No errors.
211013 13:00:11 xbcloud: Download completed.

But in reality those two folders are missing and i cant find any evidence in log that they are listed for download or any other error about download failure or connection issues.

1 Like

Hello @iadv ,

We have a healtly cluster that has backups enabled on s3 storage.

Is this cluster running in k8s? Is it deployed with the Operator?

We use https://dcd.ionos.com/

Are you using it to take the backups?

But when downloading files from s3 there are missing folders. But no errors from log.

I would love to learn more about how you take the backup first.

1 Like

The cluster is on openshift 3.11 and was deployed with operator.

app.kubernetes.io/managed-by: Helm
app.kubernetes.io/name: pxc-operator
app.kubernetes.io/version: 1.8.0
helm.sh/chart: pxc-operator-0.1.12

Yes, backups are created using built in feature
No errors there. Backups are performed successfuly.

  backup:
    image: 'percona/percona-xtradb-cluster-operator:1.8.0-pxc8.0-backup'
    imagePullPolicy: Always
    schedule:
      - keep: 72
        name: customer-db-pxc-backup-s3
        schedule: 5 * * * *
        storageName: s3-ionos
1 Like