You are not logged in.
There are some problems with update-alternatives and environment variables
Can you please elaborate on that? I highly doubt the AdoptOpenJDK people will make it depend on systemd with some minor upgrade. Of course, if you switch from openjdk8 to 11 or 14 - who knows what will happen in future? But as as long as you stick with openjdk8 I find such a change highly unlikely.
=> solve your environment troubles and enjoy a stable setup.
Hi,
I would like to tell initramfs to just continue booting. Normally, just calling "exit" does that. However, I access the system over dropbear-initramfs and thus "exit" just quits the SSH session. So - which command is it? Do I have to call init in initramfs and tell it to shut down?
Background: Due to some other trouble the cryptroot_unlock script does not properly work for me via dropbear-initramfs. However, I can call cryptsetup manually just fine, so I would like to continue booting after unlocking the device. Of course, the cryptroot_unlock script will certainly do the magic here, but I was unable yet to find out how it does that exactly.
Many thanks in advance!
Edit: I found out that "/usr/sbin/run-init" is the script to call, now need to find the correct parameters for it.
The package firmware-realtek is indeed in non-free. And yes, the installer does include non-free firmware.
See also this discussion back from 2018:
https://dev1galaxy.org/viewtopic.php?id=2475
Regarding your actual issue: I think as well that the driver may not be in 4.19. Please try the backport kernel and report back.
Well, that's basically the same approach I am taking here. The UUIDs are of course not correct since they differ between these two systems. The problem is that even after correcting the UUID (or placing PARTLABEL= there), regenerating the initramfs does not update its crypttab file as the script seems to take these parameters somewhere from the currently running system and ignores the crypttab file.
Eventually I found a way to resolve the issue by using the "cryptopts=source=..." kernel parameter. Yet I would be interested in properly unpacking/repacking the initramfs in case I'd like to manually change something there.
Another workaround is to boot the target once, bail out to initramfs prompt, mount it from there and regenerate the initramfs while booted. However, this does not work nicely if you want to prepare images for headless machines.
Thanks and yes, this my first guess as well. I tried both including gzip -9 in the pipeline (as explained in the article) and also not doing it but compressing it afterwards by hand (also with gzip -9). However, it went from ~237M uncompressed to the mentioned ~128M compressed but not down to ~27M.
Thanks, well the issue here was that I tried to move from an unencrypted to an encrypted system which resulted in a inproper initramfs.
Hi,
I am trying to port an encrypted devuan installation from system A (virtual) to system B (physical). I tried many things (e.g. using partlabels) but I can't convince the initramfs tools to generate the proper crypttab file for the initramfs.
To fix this manually, I tried to unpack and repack the initramfs as explained in this article: https://access.redhat.com/solutions/24029
It seems to work, but by just unpacking and repacking the initramfs it grows from ~27M to ~128M and fails to boot (kernel panic).
What is the proper way to unpack and repack the initramfs image if you want to do it by hand (for whatever reason)?
Thanks!
This seems to do the trick - thank you a lot!
Thank you! But I will have to install the package from unstable once for this to work, i.e. just running the dist-upgrade will not work, right?
Hi,
I tried to install the unstable kernel on beowulf (not from backports) as follows:
cat <<'EOF'>>/etc/apt/sources.list.d/unstable.list
deb http://deb.devuan.org/merged unstable main
EOF
cat <<'EOF'>>/etc/apt/preferences.d/99unstable
# Never prefer packages from unstable
Package: *
Pin: release a=unstable
Pin-Priority: 1
# Allow upgrading kernel from unstable
Package: linux-image-arm64
Pin: release a=unstable
Pin-Priority: 500
EOF
apt update
apt dist-upgrade
This used to work fine in the past, but since the new kernel 5.7 is available in unstable it tells me that packages have been held back.
When I try to install the package "linux-image-arm64" manually, it tells me that it has an unmet dependency to linux-image-5.7.0-1-arm64 that is not going to be installed. Simply installing this dependency by hand resolves the issue. However, I would like to get automatic updates when new kernels become available as well, so this is not the right solution I guess.
Any ideas how to solve this better?
It's worth a try, but I made the experience that some hardware which is "certified" for RHEL/CentOS needs some proprietary driver package provided by the vendor. And these are usually only available in RPM format and may be RHEL-specific, so I wouldn't bet all will work fine with Devuan before trying.
However, you are often better off without these proprietary drivers anyways. Not sure about this higher-end server you got.
Another issue might be that you are probably paying a lot to HPe for support, but running an "unsupported" configuration might lead to support not being available. Yet, especially for HPe servers support is necessary, as they are usually not making firmware updates available without a valid subscription which is why I avoid buying hardware from this vendor.
As an alternative, you can also consider installing AdoptOpenJDK from their repository targeted at Debian 10:
Hi,
after installing the dropbear-initramfs package, everything looked fine. However, after SSH'ing into the busybox environment, each call to
cryptroot-unlock
resulted in the following message:
Try again later
it seems like the following check in cryptroot-unlock never passes:
if [ ! -f "$TABFILE" ] || [ "$TABFILE" -ot "/proc/1" ]; then
# Too early, init-top/cryptroot hasn't finished yet
echo "Try again later" >&2
exit 1
fi
Workaround: Uncomment the line with "exit 1", then everything should work.
Is this check systemd-related?
@fsmithred: The link to your modified version stopped working, I assume this one is supposed to be the current one:
Thank you! This didn't do the trick for me - maybe because I deleted the file? Does it have to be executable?
I agree and marked the thread as solved, since the "netinst.iso" and the "mini.iso" seem to be equivalent.
To those interested in the RPI support topic: Feel free to start a new thread.
Hi,
I have an installation which initially had swap but now doesn't have it anymore. Yet, the entries for suspend to ram are effective and lead to boot delays. The following message is repeated several times:
Begin: Running /scripts/local-block ... done.
After some seconds, it finally gives up:
Have up waiting for suspend/resume device
I already tried altering and removing the file
/etc/initramfs-tools/conf.d/resume
but this didn't help.
Thanks in advance for any ideas on how to solve this.
@yeti: Thanks for the hint, I didn't know about these. Do you know from which repository they pull the kernel updates?
If Debian has a repo for that, I am sure Devuan could follow the same approach to build images for the RPI.
I see your point. However, please note that Devuan ships with an unmodified Debian kernel. The Debian kernel is a mainline LTS kernel with very few patches and *does not* support any of the RPI boards.
Given this situation, I don't see any sound way for providing any sound RPI support in Devuan. Just two ideas:
(1) Imho, the best approach would be to build some kind of "Respuan" that includes all the RPI-specific stuff but uses Devuan instead of Debian. Not sure if anyone tried this already?
(2) As an alternative, you could package a kernel for the RPI boards and then build installation images which use this kernel. However, keep ind mind that one of the great benefits of Debian's "universal OS" approach is having one generic kernel per architecture that works for all supported devices...
Nevertheless, shipping images with a kernel that is not packaged and receives no security updates whatsoever is probably not the best way to convince people about the benefits of Devuan's core philosophy - at least if you ask me.
As suggested here, I tried the "netboot mini iso" from here:
https://pkgmaster.devuan.org/devuan/dis … t/mini.iso
Interestingly, this worked like a charm - without setting up any netboot server or alike! Now I just wonder what the difference between the netboot and netinst ISO actually is.
Regarding the RPI side-topic: As long as the packaged Devuan kernel is not working for the RPI, I don't see much value in providing such images. This sort of images with a "hardcoded" kernels are definitely not what you want to run on a production system connected to any non-trustworthy network. The result of such images is that some users that are not aware of this issue, develop a false sense about security of such systems and run them in production with no kernel updates forever.
I know about this installation type, but this is just not the same. With the ISO, I can boot the ARM installer directly in Qemu and install the system comfortably using the installer like I would do for x86.
@Tatwi: Your suggestion helped with fixing the audio - I am now also getting on-screen notifications when using the hardware buttons. I am always amazed how great the Devuan community is! Without the tweaks, I would have ditched this installation right away as it had too many issues and now it runs great.
One (more?) remaining issue I still struggle with: Shutdown sometimes (but not always) takes quite long and hangs when shutting down anacron. Not sure if this is related to the disk encryption or anacron itself. Any ideas about that? Could it be another issue, such as having network manager manage the LAN connection but no network cable attached?
And even one more: Muting the audio using the hardware button does not work.
Hi,
unlike Debian, Devuan does not provide installer ISOs for architectures other than x86/x86_64. Is this intentional? Being able to install e.g. arm64 from ISO in a VM would be a huge plus for me.
Update: Adding the video module (in my case: i915) and running update-initramfs as described by @Head_on_a_Stick did the trick - many thanks! One annoyance less. Of course it would have been better if it worked ootb, but it's very nice to see that some issues can be solved with the help of the community.
I will try to fix the audio issues next.
@Head_on_a_Stick: Thank you, I will try the mentioned suggestions!
@rolfie: Yes, I saw that it's fixed in Chmaera. And by not visible I mean the following: When booting up, I only see "loading kernel .. loading initrd" and then nothing, like the system hangs. But when I enter my password, it continues to boot and I see the password prompt - but only afterwards, not when I'm actually asked to enter the password.
I use unencrypted /boot + encrypted root (not separate home) without LVM, EFI or other additional stuff. Just this dead-simple setup.