Just noticed sci-ml/ollama with a newer version of ollama. That installs without issue.

Emerge -t output:

[ebuild N ] dev-util/hip-7.1.0:0/7.1::gentoo USE="hip -debug -doc -numa (-opencl) -test" LLVM_SLOT="20" VIDEO_CARDS="nvidia -amdgpu" 0 KiB
[ebuild N ] dev-libs/rocm-core-7.1.0:0/7.1::gentoo 0 KiB
[ebuild N ] dev-libs/hipother-7.1.0:0/7.1::gentoo 0 KiB

File a bug against hip ...

I am trying to install app-misc/ollama and am coming up with file collisions between dev-util/hip-7.1.0 and dev-libs/hipother-7.1.0.

* Messages for package dev-util/hip-7.1.0:

* Detected file collision(s):
*
* /usr/include/hip/nvidia_detail/nvidia_channel_descriptor.h
* /usr/include/hip ...

It's my understanding that "rootwait" applies to the first stage of boot, while the kernel is waiting for its initial root device, which it gets. I'll try adding that and report back. rootwait is useless when used with most initramfs because the kernel has a "root" in the initramfs.

The default of ...

[...] All I have to do is load the module [...]
Which module?

I can drop into the rescue shell, modprobe sdhci_pci, and continue booting. Any advice?


Maybe you are only missing the kernel command line parameter "rootwait" ?

It's my understanding that "rootwait" applies to the first stage ...

Some very modern machines need SPI to be able to access MMC. Maybe have a look into the link of this chapter:
https://wiki.gentoo.org/wiki/User:Pietinger/Tutorials/Manual_kernel_configuration#Finalizing

Maybe boot with our GentooLiveCD and check with "lsmod" all loaded modules.

I'm pretty sure ...

Hia. I'm using dracut for initramfs with gentoo-kernel. It fails to early-load the drivers for my root drive. I can drop into the rescue shell, modprobe sdhci_pci, and continue booting. Any advice?

I asked nicely.

0 aperture /root # cat /etc/dracut.conf.d/mmc_modules.conf
add_drivers+=" mmc_core ...

Found a solution on the platformio forum. (Wow, google sucks these days.)

https://github.com/platformio/platform- ... 3155057733

Remove the $HOME/.platformio/packages/tool-esptoolpy directory and re-open the project in vscode.

I am starting out development for the ESP32 platform with platformio on vscode. When I try to build a basic project I get the following error:

Generating partitions .pio/build/lilygo-t-display-s3/partitions.bin
Traceback (most recent call last):
File "/home/brundage/.platformio/packages/tool ...

I see that now. Good eye. I'll untangle that mess too. Thanks.
Please forget all what I said ... sometimes I am an idiot ... :oops: ... I am so used to do ALWAYS a "iptables -L -vn" ... the -v is important to get also the information about the interface ... and yes I saw no interface information ...

[...] Just for completeness I checked iptables too. Guess docker uses it (Is there a way to change that?). Looks open.
Not quite ... yes, anything INTO this station is allowed -> Chain INPUT (policy ACCEPT) ... and everything out: Chain OUTPUT (policy ACCEPT) ... but look at FORWARD

0 panopt ...

brundage,

Without masquerading/NAT packets from the 10.0.3.0/24 net will have 10.0.3.0/24 addresses.
How will hosts on 10.0.0.0/24 receive them.
Forwarding will do that for you.

You will need a 10.0.3.0/24 address to actually receive them.

Yah, panopt (the camera net router) has a default route ...

I'm setting up a router following the Home router guide . It will be between my home network and an internal ip camera network so NAT is not necessary. Packets aren't traversing the router and I'm stumped.

0 panopt /root # uname -a
Linux panopt 6.12.31-gentoo-dist-hardened #1 SMP PREEMPT_DYNAMIC ...

Logs would be the first place to look

I've only seen it when there was a network configuration issue.

Link flap is a condition where a communications link alternates between up and down states.[1] Link flap can be caused by end station reboots, power-saving features, incorrect duplex ...

I recently upgraded my system. On the new hardware the Intel 82599 10 Gigabit NIC thrashes up and down every second or so. What's a good place to start diagnosing the problem?

The device is correctly identified

09:00.0 Ethernet controller: Intel Corporation 82599 10 Gigabit Network Connection ...

flexibeast wrote:Your /etc/conf.d/net doesn't show any configuration for the interface enp2s0 (i.e. the interface not accepting the offer). And have you created a net.enp2s0 symlink and enabled that service?

That's the server interface. It is correctly offering DHCP on that NIC. Edited the post to be more clear.

I'm running a multihomed dhcp client. Both interfaces solicit dhcp and the server answers both with an offer. dhcpcd never acknowledges offers on one interface. There is no firewall blocking dhcp.

The NIC not getting DHCP is enp0s31f6 .

Debug log:

0 blaze /root # /etc/init.d/dhcpcd restart ...

szatox wrote:How does 10Gb Eth look cost-wise right now though?

It still ain't that cheap, but attainable.

• $400 Router: Gowin GW-R86S-U3
• $180 Switch: Mikrotik CRS310-1G-5S-4S+IN
• $120 (3) network cards
• $200 SFP+ transceivers and patch cables

And I'm asking if there is a way to automatically prefer keepassxc. I probably won't remember why keepassxc is in my world file 18 months from now.
Not without writing your own ebuild for the virtual in an overlay changing the order of the packages.

Cool, thanks. At least I have this forum post ...

As grknight just told you, it will pick the first one in the list if none are already installed. If you install any element from the list, then it will be satisfied and will not prompt you to install other members of the list.

And I'm asking if there is a way to automatically prefer keepassxc. I ...

... but pambase and secret-service still wants to bring it in even though the flag explicitly unset:

0 blaze /root # equery d gnome-keyring
* These packages depend on gnome-keyring:
sys-auth/pambase-20240128 (gnome-keyring ? gnome-base/gnome-keyring[pam])
virtual/secret-service-0 (gnome-base ...

0 blaze /root # emerge --info
Portage 3.0.65 (python 3.12.3-final-0, default/linux/amd64/23.0/desktop, gcc-13, glibc-2.39-r6, 6.6.38-gentoo x86_64)
=================================================================
System uname: Linux-6.6.38-gentoo-x86_64-Intel-R-_Core-TM-_i5-6500_CPU_@_3.20GHz-with ...

I'm trying to get rid of gnome-keyring but portage insists on keeping it around.

I have the gnome-keyring USE flag unset:

0 blaze /root # grep -r gnome-keyring /etc/portage
/etc/portage/package.use/obliterate-gnome-keyring:sys-auth/pambase -gnome-keyring
/etc/portage/make.conf:USE="$USE -gnome ...

grknight wrote:Disable vaapi USE on media-video/vlc to resolve this. See also Bug 864721

Success. Thanks for the help.

brundage,

Usually have faster network require faster CPU, because rarely the "product" (byte and bits) are produced by network card. it is CPU that generate those bytes and bits. So do your desktop have fast enough CPU to generate? And what application(s) on your desktop that would generate this ...