access rules hanging in queued_to_apply after replica create

[Maurice Escher]

Hi,

I'm running xena with dhss=true and the netapp driver.

I have a share with an active access rule.

Now I create a replica of type DR.

Expected result: access rule stays active
Actual result: access rule is queued_to_apply

The only way to get out of this is to create any dummy access rule (and delete it).

I had a look into the database:
share_instance_access_map for the active replica shows state 'active'
share_instance_access_map for the non-active replica shows state 'queued_to_apply'

get_aggregated_access_rules_state() of manila/db/sqlalchemy/models.py is always looking at all rules, even in the replication type DR case, where the rules are not in use, yet.

Not even the startup ensure loop is correcting this situation since the access_rule_state of the non-active share instance is set to 'active' (and not 'syncing')

I'm unsure about the way out: should the get_aggregated_access_rules_state() method conditionally check? Or should the rules be active right at replica create time?

Best,
Maurice

P.S.: the state is only 'cosmetically' wrong, that means the rules stay correctly applied on the backend. But any client relying on the state will be misled.

[Maurice Escher (maurice-escher)]

Additional thoughts: even promoting a non-active replica does not help

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/manila-tempest-plugin/+/868340

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/manila/+/868341

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/manila 17.0.0.0rc1 release candidate.