Hello List,
 
i found in the meantime that the root of the problem is
that the iscsi client component of the driver is not able to connect to the iSCSI array from within the cinder-volume docker container.
But the same is possible from within the controller node, outside of the container...
 
ICMP and even SSH from within the container to the iSCSI array work perfectly fine though.
 
Therefore i believe this is most likely a sideeffect of the containerisation in the context of the Kolla-Ansible architecture...
 
Not sure yet why the driver client behaves differently from the ssh client...
 
Best
 
 
Stefan
 
Gesendet: Samstag, 20. März 2021 um 17:25 Uhr
Von: "Stefan Kelber" <Stefan.Kelber@gmx.de>
An: "Radosław Piliszek" <radoslaw.piliszek@gmail.com>, openstack-discuss@lists.openstack.org
Betreff: Aw: Re: OpenStack Victoria - Kolla-Ansible - Cinder - iSCSI-backend - Update driver status failed: config is uninitialized.
Hello List,
 
i managed to miss to reply to the list, hooray...
 
That's why yoctozepto already had a look at the attached logs/config.
His take is that it is most likely in the realm of the driver, since the client keeps attempting to connect, which fails.
He also pointed out that the driver must end up in the container,
which fortunately is s.th. already taken care of, so the attached logs reflect that state.
 
My hope is that someone recognises this error pattern from his/her iSCSI implementation,
since it seems to be quite typical...
 
Best
 
Stefan
 
Gesendet: Samstag, 20. März 2021 um 10:36 Uhr
Von: "Radosław Piliszek" <radoslaw.piliszek@gmail.com>
An: "Stefan Kelber" <Stefan.Kelber@gmx.de>
Cc: "openstack-discuss" <openstack-discuss@lists.openstack.org>
Betreff: Re: OpenStack Victoria - Kolla-Ansible - Cinder - iSCSI-backend - Update driver status failed: config is uninitialized.
Hello Stefan,

try adding in cinder-volume config, DEFAULT section:

debug = true

and then attaching its logs.
That sometimes unveils the underlying issue.

-yoctozepto

On Sat, Mar 20, 2021 at 12:41 AM Stefan Kelber <Stefan.Kelber@gmx.de> wrote:
>
> Hello List,
>
>
>
> i have trouble connecting an INFORTREND iSCSI array as external iSCSI backend to Cinder in OpenStack Victoria, deployed via Kolla-Ansible (All in One, it's still PoC).
>
>
>
> I use a volume driver provided by the manufacturer, which is listed as "completely" supported by OpenStack.
>
> And i can use it to log in to the array using the client.
>
> iSCSI session is established.
>
> Firewall etc off, as recommended.
>
> I can even start cinder-volume and as long as this service stays up, i can create volumes and remove them, but always they are reported in "Error" state.
>
> Soon after having started the service cinder-volume, the service shuts down.
>
> When reading the logfiles, it looks like the following error is the first one to show up, when starting the cinder-volume:
>
>
>
> cinder-volume.log:
>
> WARNING cinder.volume.manager [req-1290dad2-d09e-40f1-9b38-2f4b5e7accb8 - - - - -] Update driver status failed: (config name IFT-ISCSI) is uninitialized.
>
> ERROR cinder.service [-] Manager for service cinder-volume host@IFT-ISCSI is reporting problems, not sending heartbeat. Service will appear "down".
>
> DEBUG oslo_concurrency.lockutils [-] Acquired lock "singleton_lock" lock /usr/lib/python3.6/site-packages/oslo_concurrency/lockutils.py:266
>
> DEBUG oslo_concurrency.lockutils [-] Releasing lock "singleton_lock" lock /usr/lib/python3.6/site-packages/oslo_concurrency/lockutils.py:282
>
>
>
> cinder.conf is setup according to documentation by the manufacturer, and matches what i found similar from DELL and HITACHI.
>
> Googling for this error unveils that this is a very frequent error, but non of the proposed solutions works for me.
>
> Does any body have a clue where to look?
>
> I am running out of ideas, unfortunately
>
> Best
>
> SK