On my laptop, I use unified kernel images, secure boot, measured boot (systemd-cryptenroll with TPM 2.0 pin), FDE and btrfs/mdadm RAID.

My disk layout looks basically like:


PC∕Laptop──────────────────────────┐
└── ∕dev∕sda └── ∕dev∕sdb
├── 1. EFI System Partition ├── 1. EFI System Partition ...

I explained the cause for this issue in my comment at:
https://bugs.gentoo.org/824782#c17

The PR to fix this is linked at the end.

Change my name to duxsco, please. Thx

Sorry for the inconvenience. But, can you change my username to:

duxco

Thx

Please, change my username to:

avidari

Nreal wrote:How could you bond two connections - wifi and wired to one? Wired 10 mbs from router enp4s0 and wifi from adroid phone 300mbs/s.

Afaik, that's not possible. See:

https://unix.stackexchange.com/question ... 105#132105

I couldn't find a solution. If you have both a SSD and HDD, you can use the SSD hardware encryption and LUKS on HDD. Then, mirror boot, root etc. over both drives. Unfortunately, btrfs doesn't support the "--write-mostly" option, known from mdadm. Or, you patch the bootup files like astroe.

I have AES-NI and fault tolerance is more important for me than some loss of performance. If one block has gone bad, btrfs recovers by using the healthy block from the other hard drive upon read time. Afaik, mdadm doesn't store checksums on data and metadata like btrfs does (https://btrfs.wiki ...

Hi.
I wish to have the following setup:


/dev/sda1: mdadm ___
\
|---> /dev/md0 (raid1) -> ext4 -> /boot
/dev/sdb1: mdadm ___/


/dev/sda2: mdadm ___
\
|---> /dev/md1 (raid1) -> luks -> swap
/dev/sdb2: mdadm ___/

/dev/sda3: luks ---> /dev/mapper/root ---> btrfs raid 1 parition ...

mv wrote:I guess that you are looking for the emerge option --noconfmem

Thanks. That solved the problem.

I noticed the existence of "/etc/config-archive". Does the deletion of its content solve the problem?
e.g.:
1. Delete content of "/etc/config-archive"
2. Run e.g. "emerge -eav world"
3. Run "dispatch-conf"

The system tells me that there is nothing to do although some config files differ from the original ones.
Btw, I usually use dispatch-conf. But, portage should notice differing config files either way after an emerge.

Portage tells me to use etc-update, if I, for example, install apache, edit the config files and reemerge with additional apache modules.
After a while, portage ignores config files that differ from the original ones and doesn't ask for etc-update.
How can I tell portage to ask me every time?

If I ...

thanks

I executed "emerge -e world" to compile everything with "-march=native". But, I had to execute "kill -9" on emerge in order to execute "pump emerge -e world". Do I have to consider anything (e.g. some files left behind from the first emerge)?


Would you like to merge these packages? [Yes/No] yes ...

Afaik lvm devices are created in the /dev directory and a symlink is created in /dev/mapper thats point to that lvm device:

latitude ~ # ls -la /dev/mapper/vg-*
lrwxrwxrwx 1 root root 7 8. Dez 16:10 /dev/mapper/vg-root -> ../dm-4
lrwxrwxrwx 1 root root 7 8. Dez 16:10 /dev/mapper/vg-swap -> ../dm-3 ...

Both solutions solve the problem. Thanks

Hi *,

rstudio requires libxml2 to be built without the icu USE flag and chromium needs it to have this flag:


latitude ~ # emerge -av rstudio

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild N ] dev-libs/mathjax-2.2 USE="-doc -examples" 8,764 kB ...

I think you should use the larc.org tutorial and ping based fail-over in order to get a dual wan setup where the server is directly connected to the internet.

So, as soon as one "slot" of the 36 becomes free the next build should start.

Hi.
I have this in my make.conf for simultaneous emerge:

MAKEOPTS="-j38 -l36"
EMERGE_DEFAULT_OPTS="--jobs=36 --load-average=36"
FEATURES="distcc webrsync-gpg"

With these lines, I make portage build up to 36 packages at the same time. But, in order to build the next 36 packages the former 36 ...

There might be use cases where a direct connection of the server to the internet is not possible, e.g. IP-based telephony, IP-TV, restrictions of TV-cable based connections (KabelDeutschland), VDSL. The advantage of my solution is that it is applicable under more use cases. As long as you have ...

If I am not mistaken, your routing tables are like the following:

- table T1:

default via 1.0.0.254 dev eth0 proto static
1.0.0.0/24 dev eth0 proto kernel scope link src 1.0.0.10


- table T1:

default via 2.0.0.254 dev eth1 proto static
2.0.0.0/24 dev eth1 proto kernel scope link src 2.0.0.10 ...

What do you use for $P1_NET and $P2_NET?

Larc.org says: In both cases, the network is the internet. So $P1_NET == $P2_NET.

Another solution to this problem is splitting all non-private subnets up between the two wan NICs.
Very simple (and wrong) example:
-first wan gets IP ranges 0.0.0.0-127.255.255.255
-second wan gets IP ranges 128.0.0.0-255.255.255.255

But, I don't like this solution, because it's not round-robin.