You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The description of this issue and possible solutions can be found in this doc.
The problem is that sm-agent configures object storage access via scylla-manager-agent.yaml, but Scylla uses object_storage.yaml for this purpose. Those files have different formats and it's not obvious how to match object storage configurations between them.
One of the possible solutions to this problem is #4161, but we haven't yet decided if that's the way to go.
UPDATE:
We decided that Scylla will expose an endpoint which would (issue):
list object storage configurations
tell which configuration is the default
Extending location flag with endpoint instead of provider is still needed (issue).
If user runs backup/restore just with the provider specified, SM will:
use the default Scylla object storage configuration
use the Rclone configuration for specified provider
validate that both configurations match
If user runs backup/restore with specified endpoint, SM will:
use the provided endpoint Scylla object storage configuration
use Rclone configuration for matching provider
validate that both configurations match
Note that different nodes might have different configurations, so the validation needs to be per node (even though we allow for specifying provider/endpoint only per DC with --location falg)
The text was updated successfully, but these errors were encountered:
The description of this issue and possible solutions can be found in this doc.
The problem is that sm-agent configures object storage access via
scylla-manager-agent.yaml
, but Scylla usesobject_storage.yaml
for this purpose. Those files have different formats and it's not obvious how to match object storage configurations between them.One of the possible solutions to this problem is #4161, but we haven't yet decided if that's the way to go.
UPDATE:
We decided that Scylla will expose an endpoint which would (issue):
Extending location flag with endpoint instead of provider is still needed (issue).
If user runs backup/restore just with the provider specified, SM will:
If user runs backup/restore with specified endpoint, SM will:
Note that different nodes might have different configurations, so the validation needs to be per node (even though we allow for specifying provider/endpoint only per DC with --location falg)
The text was updated successfully, but these errors were encountered: