Benjamin Drung

I did a benchmark on my laptop to compare 3cpio and dracut-cpio and they have the same speed on a cold cache: ``` $ hyperfine -w 2 -r 100 -p...

dracut-cpio is slightly faster than 3cpio with a warm cache. I attribute the difference to the manifest support since 3cpio reads the metadata first before starting to construct the cpio....

Thanks for the report. We either need to use `set -o pipefail` (the big hammer) or hunt down all `&&` and `||` cases that need to be fixed.

These ugly behaviours of shell tickle my nerves. What prevent us from using a nicer scripting language like Python and a testing framework like pytest for our tests?

I am open to improve the release process. IMO the main branch should always be in an releasable state and we should use a time-based release. I am open to...

We could increase the release schedule to do one release every two month.

The amount of changes will be the same, but cut into smaller chunks, won't it? If downstreams want to invest less time they can skip upstream releases IMO.

@MorningLightMountain713 I tried to reproduce the issue with Ubuntu 25.04 in a chroot. Both as user and as root the `etc/passwd` and `etc/shadow` have mode 644. Have you changed your...

+1 for switching to dnsmasq. dnsmasq is small and easy to configure.

From the boot log: > [ 0.000000] Command line: BOOT_IMAGE=/vmlinuz-6.14.0-1007-oracle root=UUID=601d6b20-8767-4f29-8aa0-18bb11ded3e7 ro console=tty1 console=ttyS0 nvme_core.shutdown_timeout=10 crash_kexec_post_notifiers rd.luks=0 rd.md=0 rd.dm=0 rd.net.timeout.carrier=5 rd.iscsi.param=node.session.timeo.replacement_timeout=6000 net.ifnames=1 ipmi_si.tryacpi=0 ipmi_si.trydmi=0 libiscsi.debug_libiscsi_eh=1 rd.net.timeout.dhcp=10 ip=dhcp netroot=iscsi rd.iscsi.firmware=1