New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Make it possible to restore schema into a different DC setting #4049

Open

Michal-Leszczynski opened this issue Sep 30, 2024 · 2 comments

Assignees

Labels

enhancement restore

Milestone

Collaborator

Michal-Leszczynski commented Sep 30, 2024

As described in #4041 and #3525, starting from SM 3.3 and Scylla 6.0, SM restores schema by applying the output of DESC SCHEMA WITH INTERNALS. As DCs are a part of keyspace schema, it makes it impossible to restore schema into a cluster with different DCs - even though we are aiming to make restore work with any cluster topology.

So right now this is a documented limitation, but we should make it possible to restore schema into a different DC setting or make it easier for the user to modify just the DC part of keyspace schema.

cc: @karol-kokoszka @tzach @mikliapko

The text was updated successfully, but these errors were encountered:

Michal-Leszczynski added enhancement refinement-needed restore labels

Michal-Leszczynski self-assigned this

This was referenced Sep 30, 2024

After a successful schema and data restoration *to a different region*, the restored keyspace is completely empty #3525

Closed

Schema restore failed with: "Unrecognized strategy option {us-east} passed to org.apache.cassandra.locator.NetworkTopologyStrategy" #4041

Closed

mikliapko mentioned this issue

Fix restore nemesis tests to correctly handle the case when the region of the snapshots doesn't match #4052

Open

Michal-Leszczynski mentioned this issue

Allow for creating stand-alone dc filter #4061

Merged

Collaborator

karol-kokoszka commented Oct 7, 2024

See #4052 (comment)

karol-kokoszka removed the refinement-needed label

Michal-Leszczynski mentioned this issue

Give possibility for restoring DC using mapping sourceDC -> destinationDC #3829

Open

karol-kokoszka modified the milestones: backup/restore with Scylla API, 1-1 Restore

Michal-Leszczynski assigned VAveryanov8

mikliapko added a commit to mikliapko/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

36704a2

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

mikliapko mentioned this issue

Support different regions in disrupt_mgmt_restore nemesis scylladb/scylla-cluster-tests#9492

Merged

5 tasks

mikliapko added a commit to mikliapko/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

e546534

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

mikliapko added a commit to mikliapko/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

33594d2

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

mikliapko added a commit to mikliapko/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

0bc578b

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

b161890

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

53fbf07

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

371096a

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

9e23c8b

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

ab3f403

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

41e7522

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

c9030e9

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mergify bot pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

491f9ff

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

c8a5a8a

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

9fda640

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

07ea003

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

d63803d

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

d8e4d37

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

mikliapko added a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

047b143

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

47ad11f

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

7059be1

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

50a5f1d

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

d4c0d7a

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(manager): skip set ks strategy and rf for disrupt_mgmt_restore

50e343a

According to comment (1), set ks strategy and rf is not needed if
restoring the schema within one DC.

It should be brought back after implementation of (2) which will
unblock schema restore into a different DC. For now, it's possible
to restore schema only within one DC.

Refs:
#1: scylladb/scylla-manager#4041
issuecomment-2565489699
#2: scylladb/scylla-manager#4049

(cherry picked from commit b161890)

fruch pushed a commit to scylladb/scylla-cluster-tests that referenced this issue


          fix(nemesis): skip disrupt_mgmt_restore if multiDC cluster

disrupt_mgmt_restore doesn't support multiDC cluster configuration due
to the issues:
- scylladb/scylla-manager#3829
- scylladb/scylla-manager#4049

Thus, it should be skipped if both issues are opened and cluster
configuration is multiDC.

(cherry picked from commit 53fbf07)

Collaborator

VAveryanov8 commented Jan 23, 2025

We discussed this task with @karol-kokoszka and @Michal-Leszczynski and came to the following conclusions:

Scylla Manager should provide an API and CLI to read the CQL schema from the backup location.
For the 1-to-1 restore effort, Siren will use this API to read the schema, modify it by changing the DC names, and then apply it directly to the Scylla cluster.

Some additional details:

It would be beneficial to have a schema-reading API not only for 1-to-1 restores, but also for other use cases when the schema needs to be changed, e.g., migration to tablets.
It is easier for Siren to apply the schema on its own because there is no need to grant and revoke permissions for the Scylla Manager CQL user.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment