Comment 66 for bug 1797581

After a longer session on IRC this is now also for Dmitrii un-reproducible.
Lets summarize the current status for the next oen coming by:

Current ideas still are:
a) at 2G the placement of kernel/initrd is too close (as placed by pxelinux), then when the kernel unpacks it overwrites the initrd
b) maas pxe backend might add some bits in transmission that break it
=> both of the above might depend on there kernel/initrd size (which would be why some kernels/daily images are affected)

We would be back needing a case that reliable triggers outside of MAAS.
If affected read through the former comments
0. safe and attach here your kernel/initrd/xml to help reproducing it in some debuggable way
0. state the Host OS version and components libvirt/qemu/maas used
1. check if it does reproduce reliably for you (retry a few times)
2. try to do the same without PXE booting (so far always resolved the case)
   (see above for details)
3a. if you get it recreated without PXE report here how you did so
3b. if only with MAAS then try to modify sizes of the roms (comment #62)
... (per discussion)