Yes, i’m the cluster admin. My cluster is based on Centos Stream / Kolla-ansible / Wallaby. You’re right, I need to check all the logs. (/var/log/kolla/cinder for example for me) Or check in containers…. But before, I'am not sure what I am trying to do is possible, and since I am not sure of my explanations (in English), it is difficult to make myself fully understood about the problem. Thank you very much for your help Franck VEDEL
Le 13 oct. 2021 à 22:06, DHilsbos@performair.com a écrit :
Franck;
What version of OpenStack are you running? Are you the cluster administrator, or a user of the cluster?
I’m running Victoria, all tips below assume that major version.
Can you create an image backed volume outside of the instance creation process?
Do you have access to the systems running the cluster, can you review logs on the controller computers? You’re looking for the logs from the glance and cinder services. Glance’s logs should be somewhere like /var/log/glance/. I only have api.log for glance. Cinder’s should be somewhere like /var/log/cinder/. I have api.log, backup.log, scheduler.log, and volume.log.
You should also check your glance and cinder configurations. They will be at /etc/glance/glance-api.conf and /etc/cinder/cinder.conf. In the glance configuration, you’re looking for the enabled_backends line in the [DEFAULT] section. If I remember correctly, it’s values has the form <name>:<type>. The type is the interesting part. Cinder is a little more difficult. You’re still going to be looking for an enabled_backends line, in the [DEFAULT] section, but it’s value is just a name (enabled_backends = <name>). You need to locate a configuration section which matches the name ([<name>]). You’ll then be looking for a volume_driver line. Based on you response, I suspect this will be: volume_driver = cinder.volume.drivers.lvm.LVMVolumeDriver.
I believe the logs will be critical to diagnosing this issue. I suspect you’ll find the error in the cinder volume.log, though it might also be in scheduler.log, or even in the glance.log.
Thank you,
Dominic L. Hilsbos, MBA Vice President – Information Technology Perform Air International Inc. DHilsbos@PerformAir.com <mailto:DHilsbos@PerformAir.com> www.PerformAir.com <http://www.performair.com/>
From: Franck VEDEL [mailto:franck.vedel@univ-grenoble-alpes.fr] Sent: Wednesday, October 13, 2021 12:02 PM To: Dominic Hilsbos Cc: openstack-discuss@lists.openstack.org Subject: Re: Problème with image from snapshot
Hi Dominic, and thanks a lot for your help. I only see one issue with what you said, or perhaps didn't say. You are aware that it is a very good idea to sysprep --generalize a Windows instance, before making an image from it, yes? Yes yes, i did that, sys prep — generalize
Regarding OpenStack, could you tell us what glance and cinder drivers you use? i’m not sure… for cinder: LVM on a iscsi bay
Have you done other volume to image before? No, and it’s a good idea to test with a cirros instance. I will try tomorrow.
Have you verified that the image finishes creating before trying to create a VM from it? Yes
I'm not sure that snapshotting before creating an image is necessary. It's a good idea (maybe even necessary) to have the instance off when converting a volume to an image, thus, depending on your storage technology, the image might just be a snapshot. I just tried with an instance off… same problem, sam error message (Block Device Mapping is Invalid)
I've done volume to image before, but it's been a little while. I plan to do this with a Linux instance today, so I'll see if I can do it while the instance is running. Thanks a lot !! Really !!
Franck VEDEL Dép. Réseaux Informatiques & Télécoms IUT1 - Univ GRENOBLE Alpes 0476824462 Stages, Alternance, Emploi. http://www.rtgrenoble.fr <http://www.rtgrenoble.fr/>
Le 13 oct. 2021 à 17:16, <DHilsbos@performair.com <mailto:DHilsbos@performair.com>> <DHilsbos@performair.com <mailto:DHilsbos@performair.com>> a écrit :
Franck;
I only see one issue with what you said, or perhaps didn't say. You are aware that it is a very good idea to sysprep --generalize a Windows instance, before making an image from it, yes?
Regarding OpenStack, could you tell us what glance and cinder drivers you use?
Have you done other volume to image before?
Have you verified that the image finishes creating before trying to create a VM from it?
I'm not sure that snapshotting before creating an image is necessary. It's a good idea (maybe even necessary) to have the instance off when converting a volume to an image, thus, depending on your storage technology, the image might just be a snapshot.
I've done volume to image before, but it's been a little while. I plan to do this with a Linux instance today, so I'll see if I can do it while the instance is running.
Thank you,
Dominic L. Hilsbos, MBA Vice President – Information Technology Perform Air International Inc. DHilsbos@PerformAir.com <mailto:DHilsbos@PerformAir.com> www.PerformAir.com
From: Franck VEDEL [mailto:franck.vedel@univ-grenoble-alpes.fr] Sent: Wednesday, October 13, 2021 12:58 AM To: openstack-discuss Subject: Problème with image from snapshot
Hello and first sorry for my english… thanks google.
Something is wrong with what I want to do: I use Wallaby, it works very well (apart from VpnaaS, I wasted too much time this summer to make it work, without success, and the bug does not seem to be fixed).
Here is what I want to do and which does not work as I want: - With an admin account, I launch a Win10 instance from the image I created. The instance is working but it takes about 10 minutes to get Win10 up and running. I wanted to take a snapshot of this instance and then create a new image from this snapshot. And that users use this new image. I create the snapshot, I place the "--public" parameter on the new image. I'm trying to create a new instance from this snapshot with the admin account: it works. I create a new user, who has his project, and sees all the images. I try to create an instance with this new image and I get the message:
Block Device Mapping is Invalid: failed to get snapshot f12c04f2-51e7-4817-ab9b-eda63c5b9aff. (HTTP 400) (Request-ID: req-c26dab86-c25f-409a-8390-8aa0ea8fe1cb)
Is it a legal problem? Is it possible to do as I do? otherwise how should we do it?
Thanks if you have ideas for helping me
Franck VEDEL