Description: When deploying an Ubuntu environment on custom hardware the LVM volumes won't get created succesfully. Note when using CentOS environment on the same hardware it works as it should.
The attachment sosreport-node-39 includes all the system information of node that has the problematic hardware.
After first using CentOS to deploy the nodes, which clears the previous disk partitions and then deploy ubuntu it's working fine. So it seems this issue is also specific to existing partitioning and ubuntu not able to cleanly delete previous existing partitioning.
Fuel version: 5.1
build_id: "2014-09-17_21-40-34"
build_number: "11"
Steps to reproduce:
1. Use same hardware environment as sos-report
2. Deploy Ubuntu environment
Actual results:
Deployment failed because it couldn't create the LVM volumes
Expected results:
Deployment succeeds for both CentOS and Ubuntu
Interesting log to see of fuel master is /var/log/remote/10.10.10.24/root.log
Sam,
could you please provide a diagnostic snapshot which is not touched? That one you attached to this bug does not contain logs for node-39 and looks like someone worked with it and removed some of the files.
What we really need to know is what was the working flow. Afaiu, if you do not install anything and then try ubuntu, it fails (because of LVM), but if you deploy centos and then again ubuntu, it works file. Right?
But we erase every node every time we are going to install OS on it. We do it before reboot, inside bootstrap OS. We write 10M from /dev/zero in the beginning of every partition and every disk. https:/
Really need untouched logs.