Differeence between in-place restore-cluster and restore-node with option --table ks.ts1 from the same diff. backup

[m934030039]

Project board link

Hi,

Any difference results between below ?

For example, we have 10 nodes (3 seed nodes,RF=3) with diff. backup to S3, periodicly for 10 days.

To perform on one node with pssh utility.

medusa restore-cluster --table ks.ts1 --temp-dir /backup/restore --backup-name s3.diff_restore_testing --in-place

To perform node by node one at a time.

medusa restore-node --table ks.ts1 --temp-dir /backup/restore --backup-name s3.diff_restore_testing --in-place

Theoretically, we expect both of results should be the same.

In reality, the case #1 is ok but something wrong on the case #2.

We are looking into this.

We are using Medusa version 0.11.x

-Sheng

┆Issue is synchronized with this Jira Task by Unito ┆Issue Number: K8SSAND-1647 ┆priority: Medium ┆Link To Issue: https://k8ssandra.atlassian.net/browse/K8SSAND-1647

[adejanovski]

restore-cluster performs orchestration that pssh won't. If you can provide some logs on what's going wrong precisely, it would be helpful to investigate this.

[nguyenthai0107]

Hello @m934030039 Have a nice day. Just a question for you with your cluster cassandra. So when run backup. did you run backups on 3 seeds hosts or just only 1 seed host. Thank you.

[m934030039]

Hi,

All nodes includes seed nodes.

-Sheng

On Wed, Aug 30, 2023 at 11:03 nguyenthai0107 @.***> wrote:

Hello @m934030039 https://github.com/m934030039 Have a nice day. Just a question for you with your cluster cassandra. So when run backup. did you run backups on 3 seeds hosts or just only 1 seed host. Thank you.

— Reply to this email directly, view it on GitHub https://github.com/thelastpickle/cassandra-medusa/issues/496#issuecomment-1698418062, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHOILWQTVG4A2PNW6FCR7ULXX2URJANCNFSM53NIR5DQ . You are receiving this because you were mentioned.Message ID: @.***>