Open Stack

Hello everyone,

I have recently started to build new deploy images for Ironic in order not to use the default ones (will need to customize them in a future, so I want to develop a proper procedure in this area). Apparently I'm struggling with an issue in ironic-python-agent (stable/ocata) related to exposing the drive using iSCSI during deployment process.

A physical console shows the following error: "Unsupported SCSI Opcode 0xa3, sending CHECK_CONDITION"

Whereas at the same time journal log for ironic-python-agent looks as follows

INFO ironic_python_agent.extensions.iscsi [-] Created iSCSI target with iqn iqn.2008-10.org.openstack:dcf9e950-2ff7-41b2-a9ef-a589b5d5a39b, portal port 3260, on device /dev/sdb using linux-io
INFO root [-] Command iscsi.start_iscsi_target completed: Command name: start_iscsi_target, params: {'iqn': 'iqn.2008-10.org.openstack:dcf9e950-2ff7-41b2-a9ef-a589b5d5a39b', 'portal_port': 3260, 'wipe_disk_metadata': True}, status: SUCCEEDED, result: {'iscsi_target_iqn': 'iqn.2008-10.org.openstack:dcf9e950-2ff7-41b2-a9ef-a589b5d5a39b'}.

Now a little bit about deploy image I'm using -- "disk-image-create fedora ironic-agent local-config disable-selinux" with ironic-agent element modified in order to use stable/ocata branch (source-repository-ironic-agent file). I haven't modified any code, especially the one related to ironic-agent.

What really concerns me in here is when I use a deploy image provided by devstack by default -- ir-deploy-pxe_ipmitool -- everything works correctly and the user image is correctly written into the device. It means there is something different between this one and the one I have built in-house, but I don't see anything obvious what may cause the problem.

I'm facing exactly the same behaviour using stable/newton (for all the components = devstack + deploy image).

Has anyone ever seen a similar issue and/or can point me what else can I look for?

Thanks for help,
Mateusz