Conversation
`--fstab` shows only mounts defined in `/etc/fstab`, which however should be fine. The benefit is that it shows also drives which are not currently mounted, or not mounted yet as of `x-systemd.automount`. This also removes the unnecessary `-D` option from the other call. An issue which still needs to be addressed, is to trigger `x-systemd.automount`s before obtaining usage stats. Otherwise no-yet mounted drives from patterns are missing, and explicit selections will show `???` placeholder values. A cheap/efficient but effective method to `test -e` or similar matching mountpoints, before the `findmnt` call, needs to be found.
It is compiled against qpdf from Raspbian, which seems to be ABI-incompatible with the one from Debian. The only RPi repo patch is preventing the `cups-browsed` service from the same-named package from being enabled and started on package install. Something which is not relevant for DietPi.
425 MiB only for WiFi firmware now, 70% of the overall disk usage, vs 303 MiB on Trixie ... I think we should switch to armbian-firmware, where our (reduced) builds are only 145 MiB, a much more selective set of firmware needed for onboard chips, M.2 modules, and USB dongles used on SBCs.
ChallengeResponseAuthentication appears as well in several comment lines. Just replace any occurance with KbdInteractiveAuthentication, to avoid confusion. Duplicates wouldn't cause issues either. Other thing: Recognise the Raspberry Pi Compute Modul Zero. It uses the BCM2710 SoC, and has optional onboard WiFi. Need to check for other compute modules, whether we should set G_HW_ONBOARD_WIFI=1 there as well, so that onboard WiFi can be properly toggled, if present. Need to check as well whether it causes any issue, if onboard WiFi absent.
Apply also retrospectively, since "--no-sandbox" has security impacts, should hence not be used if it is not really the root user explicitly starting Chromium.
superseded in v10.7
Also, the VirtualBox OVF did use %M where %m was intended, causing unintended and potentially invalid last state change timestamps (with however no real practical issues).
…tion (#8248) and adjust ban messages accordingly ("IPs" => "bans")
* Correctly catch sudo error in Fail2Ban status, via PIPESTATUS[0], otherwise the exit code of "mawk" is checked. * Use sudo as well for systemd status as non-root, if dbus is not active, needed for non-root users to call systemctl.
According to an RPi engineer, this is not needed on RPi 5 anymore for proper 4k support. dietpi-globals: fix comment typo and print error to STDERR
Our dietpi-build.service controlling terminal is sometimes stolen by console-getty.service, which is Type=idle as well, causing an immediate shutdown. If ordering works between Type=idle services, and console-getty (which is Before=getty.target > multi-user.target > graphical.target) claims /dev/console before follow-up targets are reached, this should reliably solve the issue. Otherwise a mask, as used before the general /dev/console issue on QEMU-emulated Forky systems was solved, solves it reliably. But late ordering of our service makes sense anyway to avoid mixing outputs etc, and we want to keep interference minimal.
It seems the more natural order, with the more important notifications right above the command prompt.
Otherwise systemd somehow seems an ordering cycle between dietpi-build.service, graphical.target, and multi-user.target. It seems, since the service is pulled in by multi-user.target (`/etc/systemd/system/multi-user.target.wants/dietpi-build.service`), this implies Before=multi-user.target, or so. Anyway, there is nothing between those two targets on a fresh Debian, hence it shouldn't make a difference. The `/etc/systemd/system/graphical.target.wants` dir does not exist OOTB, and we want to add as few stuff as possible.
Even when starting earlier, console-getty.service can take longer for the login prompt, after dietpi-build.service started, stealing its controlling terminal. Hence there is no way around masking it. The After=multi-user.target still makes sense for a cleaner output ordering. Also, disable wpa_supplicant.service before booting, as it is doomed to fail, causing an ugly failure log, and is otherwise disabled by dietpi-installer anyway.
The additional ID list here has been forgotten, and the NanoPi Zero2 (ID 104) mainline Linux builds were overriding the vendor Linux builds, due to the missing suffix. The conditional here is redundant. Our CI uses the "--rk35xx-mainline" flag only where it actually expects a dedicated image.
The notifications app (a vendored dependency of it) started to throw an info log message whenever it is called, that it "strongly recommends" GMP or BCMath for better performance. GMP is needed for the SFTP storage backend as well, and recommended when using WebAuthn authentication. And it seems to perform better. So we're going with that one.
The overlay name which sets it to host mode is different in mainline Linux.
The Kodi 20 package from the RPi Bookworm repo have a different file distribution between kodi and kodi-bin, which causes a conflict when dpkg tries to unpack files of the new kodi-bin package while the old kodi package is still installed. Purge it first, leaving dependencies in place, and let `apt-get dist-upgrade` reinstall it along with the other upgrades.
Beta v10.7.0
They were just left stale on the FAT partition after being copied to the rootfs /boot dir. Remove them after copy, and also remove them retrospectively on DietPi update, since they can contain sensitive data.
The whole code block ran only if no overlay was applied yet, but did then loop through overlays already applied. Generally it makes sense to apply the 8GB an A12 overlays where needed as well if there are already other overlays applied. Remove the condition, and set a reboot flag instead, to avoid a reboot (and boot loop) if no changes were done.
and add support for newly added Orange Pi boards
It was fixed for Odroids, while support for new Orange Pis was added.
systemd-timesyncd sometimes gives up for at least ConnectionRetrySec, before retrying to connect to time servers, e.g. if network connectivity has not yet been fully established, on its first attempt. Contradictorily, this has been especially observed when syncing time with the local gateway/router. For some reason, in my personal home network case, even that ConnectionRetrySec is 30 seconds (the default), it times out the full 60 seconds during boot, while a systemd-timesyncd.service restart results in an almost instant sync from second 1 (of our script's loop) on. This might be an edge case, probably related to some special reply of the router's time server when it is just about recognizing the peer and its network state. But in any case, it is much more likely to speed up time sync than aborting an actual sync attempt. In the latter case, the delay would be usually only a fraction of a second.
It is the same situation as with "mount", mawk, and others: "required", but not "essential", and no dependency of any other installed package. While "required" packages are not autoremoved, we better mark them as manually installed, to make clear we need/expect them.
Current container images ship without a network stack, which leaves guests with their own network namespace, like Proxmox LXC, offline: #5204 With the flag set, dietpi-build and dietpi-installer add Dropbear, ifupdown and the DHCP client the scripts already choose to the container package set, write the interfaces base config and enable the network wait at boot. First boot needs no special casing anymore: its package gates apply the network setup and regenerate the Dropbear host keys, and dietpi-network defaults containers to "auto". Guest network containers get Dropbear pre-installed and marked as installed, like regular systems. ___________ * dietpi-build: skip DHCP client if no network stack is installed After isc-dhcp-client was removed from Forky, so that we needed to switch to dhcpcd there, it was accidentally installed on container images as well, though autoremoved again in dietpi-installer. Prevent it from being installed in the first place, if no network stack is installed. Add some comments about which packages are needed on which images, an empty lines as code structure * dietpi-installer: deduplicate network stack package assignments * dietpi-build: install hdparm in dietpi-build already * dietpi-build: give networked container images a suffix and add to ci * changelog: add entry about networked container images Co-authored-by: MichaIng <micha@dietpi.com>
No idea why it was this insane 700 MiB. Final disk usage is ~270 MiB, hence 300 MiB should be sufficient.
Signed-off-by: MichaIng <micha@dietpi.com>
Co-authored-by: MichaIng <micha@dietpi.com>
* Check passed upload script existent/executability right at the start of each script * Do make actually use of the passed script path, rather than hardcoded ."/upload.sh" * The upload script is not passed to dietpi-imager if an individual VM image/appliance was requested, since the raw image servers solely as basis for qemu-img to generate the virtual disk image. But adjust logic so that the passed upload script is used for uploads of the final VM image/appliance. * Do not rename Proxmox image anymore. The QCOW2 image can be used with various other virtualizers, and the file extension is sufficient to indicate what it is. Rename the variant option accordingly. * Do not xz-compress QCOW2 images anymore. It reduces the size by a few MiB only, since we use native QCOW2 compression already. That saves users from doing an extra step. * Single-quote workflow expressions in shell context, and add ARMv8 VMs to CI
All rockchip-rk3568-* overlays start with rockchip-rk3568-hk-*, and we use the latter in all other cases.
…tomation platform (#8277) Co-authored-by: MichaIng <micha@dietpi.com>
Co-authored-by: MichaIng <micha@dietpi.com>
Signed-off-by: MichaIng <micha@dietpi.com>
systemd-timesyncd does not remove it on startup.
Beta v10.7.1
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.