Storwize SVC driver takes too long time to initialize

[Margarita Shakhova]

When we restart cinder-volume service initialization takes too long time (up to 10-15 mintues in case of 300+ LUNs on storage). This is because of ensure_export is called for each volume in 'attached' state and performs a separate lsvdisk request for each LUN.
At the same time cinder-volume service is ready to perform other RPC requests e.g. volume creation/deletion. All such requests are failed.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/546489

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/546489
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=d5f79c52d886e962e1e42af4d892b35eec5bb81f
Submitter: Zuul
Branch: master

commit d5f79c52d886e962e1e42af4d892b35eec5bb81f
Author: Margarita Shakhova <email address hidden>
Date: Wed Feb 21 07:50:47 2018 +0300

    Storwize: get list of all volumes for ensure_export.

    This patch solves the problem of too slow initialization of
    cinder-volume service.
    Previously lsvdisk() was called separately for each 'in-use' volume
    in order to check if the volume exists on the storage.
    Now lsvdisk() is called once per pool.

    Change-Id: Ia84afc12a30ea7b714b287844e81ba02ce4b0f3d
    Closes-Bug: #1749687

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/cinder 13.0.0.0b3 development milestone.