GitHub user RyanP-Testing added a comment to the discussion: Ubuntu 20.04 
Template Problem

Hey,

Based on the details provided, I would not treat this as an Ubuntu 20.04/ext4 
issue alone.

The most useful finding here is the comparison between the two VMware clusters: 
the same CloudStack template works with ESXi 8.0.0 but becomes corrupted with 
ESXi 8.0.3. You also mentioned that the problem can be reproduced with other 
templates, which points more toward the VMware/CloudStack template deployment 
path than a specific Ubuntu template.

I would suggest one simple isolation test before rebuilding the template again:

* Take the same template/image and deploy it directly through vCenter/ESXi on 
the affected 8.0.3 host.
* If it is already corrupted there, the issue is occurring during the VMware 
template/clone/storage operation.
* If it works correctly when deployed directly through VMware but fails only 
through CloudStack, then the CloudStack VMware deployment path should be 
investigated.

The `fsck.mode=force fsck.repair=yes` and `depmod -a` workaround is worth 
trying because a similar VMware-related filesystem issue was previously 
addressed this way in CloudStack. However, it does not appear to solve this 
particular case, since you already tested both and the filesystem was still 
corrupted.

I would therefore collect the following from a failed deployment:

1. CloudStack management-server log covering the deployment.
2. ESXi `vmkernel.log`, `hostd.log` and the VM `vmware.log`.
3. The exact ESXi 8.0.3 build number.
4. Datastore/VMFS version and storage configuration.
5. VMDK/controller type and VM hardware version.
6. The exact EXT4 errors from the first boot.

If the VMDK is already corrupted when the intermediate VMware template is 
created, as reported here, that is the most important clue and should be 
included in a CloudStack bug report.

For production, I would avoid deploying this template on the affected ESXi 
8.0.3 hosts until the storage/template conversion path is isolated. If ESXi 
8.0.0 is still available and stable, using it as a temporary workaround is 
safer than repeatedly repairing newly deployed filesystems.


Regards,

Ryan P.
Cloud & DevOps Engineer
AccuWeb.Cloud

GitHub link: 
https://github.com/apache/cloudstack/discussions/12610#discussioncomment-18106697

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to