PXE Installation of RHOCP4.6 fails randomly with error "premature end of input data at offset 0"

Issue

• PXE installation of Red Hat OpenShift Container Platform 4.6 fails with the error below.

[   24.696289] localhost systemd[1]: Started dracut initqueue hook.
[   24.717662] localhost systemd[1]: Starting Acquire live PXE rootfs image...
[   24.737488] localhost systemd[1]: Reached target Remote File Systems (Pre).
[   24.757763] localhost systemd[1]: Reached target Remote File Systems.
[   24.764422] localhost coreos-livepxe-rootfs[1723]: Fetching rootfs image from http://192.168.0.254/rhcos/4.6/rhcos-4.6.1-x86_64-live-rootfs.x86_64.img...
[   24.799932] localhost systemd[1]: Starting dracut pre-mount hook...
[   24.827611] localhost systemd[1]: Started dracut pre-mount hook.
[   27.040750] localhost coreos-livepxe-rootfs[1723]: Error: premature end of input data at offset 0
[   27.051071] localhost coreos-livepxe-rootfs[1723]: Couldn't fetch, verify, and unpack image specified by coreos.live.rootfs_url=
[   27.051071] localhost coreos-livepxe-rootfs[1723]: Check that the URL is correct and that the rootfs version matches the initramfs.
[   27.060982] localhost systemd[1]: coreos-livepxe-rootfs.service: Main process exited, code=exited, status=1/FAILURE
[   27.081077] localhost systemd[1]: coreos-livepxe-rootfs.service: Failed with result 'exit-code'.
[   27.081108] localhost systemd[1]: Failed to start Acquire live PXE rootfs image.
[   27.130872] localhost systemd[1]: Dependency failed for Initrd Root File System.

• The menuentry lines are as follows.

menuentry 'testworker' --class fedora --class gnu-linux --class gnu --class os {
  linuxefi rhcos/4.6/rhcos-4.6.1-x86_64-live-kernel-x86_64 nomodeset rd.neednet=1 coreos.inst.insecure coreos.live.rootfs_url=http://192.168.0.254/rhcos/4.6/rhcos-4.6.1-x86_64-live-rootfs.x86_64.img coreos.inst=yes coreos.inst.install_dev=nvme0n1 coreos.inst.image_url=http://192.168.0.254/rhcos/4.6/rhcos-4.6.1-x86_64-metal.x86_64.raw.gz coreos.inst.ignition_url=http://192.168.0.254/ignition/worker.ign ip=dhcp
  initrdefi rhcos/4.6/rhcos-4.6.1-x86_64-live-initramfs.x86_64.img
}

• Upon few tries, installation works again, what is the root cause of such failures?
• With bonding option LACP Active-Active, the failure rate is almost 50%, whereas with LACP Active-passive, the failure becomes zero.

ip=192.168.0.100::192.168.0.254:255.255.255.0:testcluster:bond0:none bond=bond0:ens3f0,ens7f0:lacp_rate=1,miimon=100,mode=802.3ad,xmit_hash_policy=layer3+4 nameserver=192.168.0.254