Reduce ESP partition size in deployment image

[cityofships]

It does not fit into 550MiB partition created by Ironic Python Agent. This is relevant when provisioning software RAID UEFI enabled servers.

https://github.com/openstack/ironic-python-agent/blob/stable/yoga/ironic_python_agent/raid_utils.py#L30

[markgoddard]

Do we need new host images with this change included?

[cityofships]

Do we need new host images with this change included?

Yes. Before that's available, one can either virt-customize local image or use IPA with ESP_SIZE_MIB modified to something bigger.

[cityofships]

Building in: https://github.com/stackhpc/stackhpc-kayobe-config/actions/runs/6495483910

[priteau]

Is this an upstream issue too? DIB defaults to 550 MiB and IPA too, but the DIB image doesn't fit into what IPA creates? Maybe IPA should be just slightly bigger?

[mnasiadka]

Is this an upstream issue too? DIB defaults to 550 MiB and IPA too, but the DIB image doesn't fit into what IPA creates? Maybe IPA should be just slightly bigger?

I think it's rather the issue of our layout that we pass to DIB?

[mnasiadka]

I'll rebuild the images as part of #721

[cityofships]

Is this an upstream issue too? DIB defaults to 550 MiB and IPA too, but the DIB image doesn't fit into what IPA creates? Maybe IPA should be just slightly bigger?

I think it's rather the issue of our layout that we pass to DIB?

It's probably rounding / calculations error, I saw the ESP partition 0,1MB too big for MD created by IPA. It's easier to change here than a hard-coded value in IPA for us.

[mnasiadka]

Merging as this is not tested in aio