clustervision
osgrab corrupted the directory ownerships?
SSSD crashed on a redeployed node based on a grabbed image from a working node. I found that the directory ownerships were changed after performing an osgrab.
In the screenshot below:
The system on the left is the default image (Rocky9.5).
The system on the right is the one deployed using the same image after an osgrab.
I performed the same test on another fresh installation of TrinityX 15 (main branch) on a Rocky 9.5 system and reproduce the same problem that osgrab modifies the ownership of directories for certain essential services.
It seems that when the image is grabbed from a node with a different set of UID/GID mappings, the directories adapt to the UID/GID of the corresponding services on the host system.
I'm not sure if this issue is specific to my setup or if it’s a more general problem.
To work around it, I mapped the UID/GID in the passwd, passwd-, group, and group- files on the resultant image (after an osgrab) to match the set from the host. So far, this approach seems to be working.
However, if this is a general problem, a better solution would be to map the host UID/GID directly during the image creation process.
we're looking into replicating the issue. So far we were not able to see the exact same issues you're seeing. It looks like rsync tries to resolve ID-s and later reverse resolve these again, which is not standard rsync behavior. Are you using a different approach for DAP backend, e.g. AD or any other setup than the default openldap configuration?
-Antoine
Please go ahead and update luna on your controllers by: ansible-playbook controller.yml --tags=luna https://docs.clustervision.com/install/upgrading/