Unable to set a custom iscsi interface

[Jack Lu]

1. create a new iface via the iscsiadm -m iface command
2. configure the iscsi_iface value to the new iface

iSCSI connector will ignore the value and will use the default iface.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/510776

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/510776
Committed: https://git.openstack.org/cgit/openstack/os-brick/commit/?id=cfe5b19fad14857773b27359eec9a8d62cf1ee7d
Submitter: Zuul
Branch: master

commit cfe5b19fad14857773b27359eec9a8d62cf1ee7d
Author: Jack Lu <email address hidden>
Date: Tue Oct 10 15:04:57 2017 +0800

    Update supported transports for iscsi connector

    In the case that a compute host with multiple NICs can be connectted to
    iscsi storage network, we expect the storage traffic only be
    transimitted via storage NIC. This need to define an custom iface and set the
    iface.net_ifacename value to the storage interface name.

    Unfortunately, the current iscsi connector does not support to use a
    custom iface. The _validate_iface_transport will change it to default.
    After add tcp to the supported_transports, it is possible to use a
    custom iface.

    Please get the details as below:
    1. CMP001 has three NICs: NIC1 is for management network, NIC2 is
    for tenant network, and NIC3 is for storage network. NIC3 can access
    storage device via layer 2 network. And NIC1 can access storage device
    too, but it can only access via layer 3 network. We hope all the storage
    traffic only be transimitted via NIC3, because they are in the same
    layer 2 network.
    2. Get the NIC3 MAC:
       fa:16:3e:79:fd:63
    3. Add new iscsi iface:
       # iscsiadm -m iface --op=new -I tcp.fa:16:3e:79:fd:63
       # iscsiadm -m iface -I tcp.fa:16:3e:79:fd:63 --op=update
                  -n iface.hwaddress -v fa:16:3e:79:fd:63
    4. Edit nova.conf, and change iscsi_iface value:
       iscsi_iface = tcp.fa:16:3e:79:fd:63
    5. Restart nova compute service:
       # systemctl restart openstack-nova-compute

    Without this change, _validate_iface_transport will change the
    custom iface tcp.fa:16:3e:79:fd:63 to default due to it cannot find tcp
    transport.

    According to this change, we should have a corresponding change to the
    nova docs, here is change to nova
    https://review.openstack.org/#/c/524443/

    Change-Id: I85a62cb6e7f8d8982e97e792d647d38ce5641060
    Closes-Bug:#1722432

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/os-brick 2.3.0 release.