cannot create access rule with non-active replica in error

[Maurice Escher]

Hi,

in a replication setup - let's assume I have one active replica B in Availability Zone (AZ) B and one non-active replica A in AZ A - my non-active replica A is in error state.

The reasons for that may be various, especially it could be that I previously had the reverse situation (active in error, something went wrong on the source AZ A) and did a replica promote to get out of that. Now I want to add new access rules, e.g. I spun up new application servers in AZ B that now need to be able to access replica B.

But this fails with
'New access rules cannot be applied while the share or any of its replicas or migration copies lacks a valid host or is in an invalid state.'

Apparently all replicas in a replication setup need to be in a non-transitional, non-error state to add new access rules.
I believe this is for historic reasons from times before share replication was implemented - git blame shows this has been added 6 years ago https://opendev.org/openstack/manila/commit/64a73b1419953bc4fbae5c1a339ad036468d54ab

Can this be adjusted?

Or is the way forward to delete the replica in error?
But since something is wrong on the backend of the replica in error, chances are high that it may not work. And force-delete is not my favourite: because either I don't allow my customers to force-delete anyhow, so they can't help themselves or it may leave orphans on the storage backend.

Thanks in advance for your thoughts,
Maurice

[Vida Haririan (vhariria)]

See additional related discussions at https://meetings.opendev.org/meetings/manila/2022/manila.2022-03-24-15.00.log.html

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/manila/+/865256

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/manila-tempest-plugin/+/865603

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/manila/+/865256
Committed: https://opendev.org/openstack/manila/commit/0282f3fdb2c0e78fa83c4b88ebe47279736cbdf4
Submitter: "Zuul (22348)"
Branch: master

commit 0282f3fdb2c0e78fa83c4b88ebe47279736cbdf4
Author: Kiran Pawar <email address hidden>
Date: Tue Nov 22 09:47:21 2022 +0000

    Ignore replicas in error state during allow/deny access.

    During replication, at least one share backed is in working and serving
    the share. So update the check of instance state from (non-transitional
    + non-error) states to only non-transitional states.

    Closes-bug: #1965561
    Change-Id: Id23d04f8d18a0e30b511c4d501a1910f2f2b4ca6

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/manila-tempest-plugin/+/865603
Committed: https://opendev.org/openstack/manila-tempest-plugin/commit/e46e2a6e415e5817ac498fead6581564d2da37b1
Submitter: "Zuul (22348)"
Branch: master

commit e46e2a6e415e5817ac498fead6581564d2da37b1
Author: Kiran Pawar <email address hidden>
Date: Thu Nov 24 20:41:07 2022 +0000

    Ignore replicas in error state during allow/deny access.

    Allow/deny access will be blocked only if share or its replica will be
    in invalid state i.e. transitional states. Error state will be ignored.

    Related-bug: #1965561
    Change-Id: I180061a50fec8a9e76ef7ceb2787239ff82a2e12
    Depends-on: Id23d04f8d18a0e30b511c4d501a1910f2f2b4ca6

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/manila 16.0.0.0rc1 release candidate.