Often getting a new hard-drive can help speed up your system, particularly if you are still using traditional platter Hard Disk Drives (HDD) and upgrade to a Solid State Drive (SSD). You will also see gains if you go from a SATA SSD to an M.2 SSD which is a newer, and faster, connection type.
After installing the new drive on to the motherboard and rebooting you should be able to find the new device. Mine was located at /dev/nvme0n1. Initially I tried the native Linux GPT partition table, but found that when writing the kernel image and initramfs I was told they could not be embedded, and rather than spend time going down another rabbit I hole I opted to switch to the more traditional MBR partition type I've used for years. Using fdisk I set the partition type to dos
fdisk /dev/nvme01n1 Command (m for help): o # Creates dos partition Command (m for help): w # Writes to disk
I then used cfdisk to create three partitions, making the boot sector bootable.
# fdisk -l /dev/nvme0n1 Disk /dev/nvme0n1: 232.9 GiB, 250059350016 bytes, 488397168 sectors Disk model: Samsung SSD 970 EVO 250GB Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: dos Disk identifier: 0xd4db420e Device Boot Start End Sectors Size Id Type /dev/nvme0n1p1 2048 167774207 167772160 80G 83 Linux /dev/nvme0n1p2 * 167774208 168183807 409600 200M 83 Linux /dev/nvme0n1p3 168183808 488397167 320213360 152.7G 83 Linux
Make file systems for each, I opted to stick with Ext4…
mkfs.ext4 -b 1024 /dev/nvme0n1p1 mkfs.ext4 -b 1024 /dev/nvme0n1p2 mkfs.ext4 -b 1024 /dev/nvme0n1p3
Mount the partitions…
mkdir -p /mnt/new-ssd/boot mkdir /mnt/new-ssd/home mount /dev/nvme0n1p1 /mnt/new-ssd/ mount /dev/nvme0n1p2 /mnt/new-ssd/boot mount /dev/nvme0n1p3 /mnt/new-ssd/home
…and copy the relevant partitions over using rsync which will preserve symbolic links…
rsync -av /boot/* /mnt/new-ssd/boot/. rsync -av /home/* /mnt/new-ssd/home/.
For the root partition I opted to exclude a number of directories using an exclude file (a simple text file that lists directory and file paths to exclude). My exclusions were in the file /root/tmp/root_fs_copy.exclude
/usr/portage /var/tmp/portage /mnt/usb* /mnt/work/* /mnt/music/* /mnt/video/* /mnt/pics/* /mnt/data/* /mnt/scans/* /boot /proc/kcore /run/* /lost+found /sys/* /home /export/* /mnt/personal/* /tmp/* /var/spool/clientmqueue/* /var/nullmailerl/* /var/lib/docker/devicemapper
To use an exclude file simply refer to it with the --exclude-from flag…
rsync -av / /mnt/new-ssd --exclude-from '/root/tmp/root_fs_copy.exclude'
Now chroot into the new environment you've just copied over. First mount /proc, /sys and /dev in the new environment.
mount --types proc /proc /mnt/new-ssd/proc mount --rbind /sys /mnt/new-ssd/sys mount --make-rslave /mnt/new-ssd/sys mount --rbind /dev /mnt/new-ssd/dev mount --make-rslave /mnt/new-ssd/dev
Then chroot into it…
chroot /mnt/gentoo /bin/bash source /etc/profile export PS1="(chroot) ${PS1}"
Finally you can install GRUB to the MBR using…
grub2-mkconfig -o /boot/grub/grub.cfg grub2-install --boot-directory=/boot /dev/nvme0n1
Update /etc/fstab to reflect the new partitions
#LABEL=root / ext4 noatime 0 1 #LABEL=boot /boot ext4 noatime,rw,users 1 2 #LABEL=home /home ext4 noatime,users,exec 0 4 /dev/nvme0n1p1 / ext4 noatime 0 1 /dev/nvme0n1p2 /boot ext4 noatime,rw,users 1 2 /dev/nvme0n1p3 /home ext4 noatime,users,exec 0 4
Exit the chroot environment, unmount all drives and reboot
exit umount /mnt/new-ssd/proc /mnt/new-ssd/sys /mnt/new-ssd/dev umount /mnt/new-ssd/boot /mnt/new-ssd/home umount /mnt/new-ssd reboot
On rebooting hit F2 or Del to enter your computers BIOS and select the new drive as the primary/first boot device. Save and exit and you should now boot into your system from the new drive.
I use UUID or LABEL to mount partitions so needed to relabel my partitions and update my /etc/fstab. Before I had my device booting and running from partitions on /dev/sda that had the following partitions and labels
# fdisk -l /dev/sda Disk /dev/sda: 232.9 GiB, 250059350016 bytes, 488397168 sectors Disk model: Samsung SSD 840 Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: dos Disk identifier: 0x75ae4ae9 Device Boot Start End Sectors Size Id Type LABEL /dev/sda1 63 80003699 80003637 38.2G 83 Linux root /dev/sda2 80003700 96004439 16000740 7.6G 82 Linux swap / Solaris /dev/sda3 * 96004440 96197219 192780 94.1M 83 Linux boot /dev/sda4 96197220 488397167 392199948 187G 5 Extended /dev/sda5 96197283 116198144 20000862 9.6G 83 Linux portage /dev/sda6 116198208 488397167 372198960 177.5G 83 Linux home
These were mounted using entries under /etc/fstab…
LABEL=root / ext4 noatime 0 1 LABEL=boot /boot ext4 noatime,rw,users 1 2 LABEL=home /home ext4 noatime,users,exec 0 4
To make it clear what these are I pre-fixed each label with sda- using e2label
e2label /dev/sda1 'sda-root' e2label /dev/sda3 'sda-boot' e2label /dev/sda5 'sda-portage' e2label /dev/sda6 'sda-home' mkdir -p /mnt/sda/root /mnt/sda/boot /mnt/sda/home
I then labelled the new partitions on /dev/nvme0n1
e2label /dev/nvme0n1p1 'root' e2label /dev/nvme0n1p2 'boot' e2label /dev/nvme0n1p3 'home'
…and updated /etc/fstab to reflect these new labels…
LABEL=sda-root /mnt/sda/root ext4 noatime 0 1 LABEL=sda-boot /boot ext4 noatime,rw,users 1 2 LABEL=sda-home /mnt/sda/home ext4 noatime,users,exec 0 4 #/dev/nvme0n1p1 / ext4 noatime 0 1 #/dev/nvme0n1p2 /boot ext4 noatime,rw,users 1 2 #/dev/nvme0n1p3 /home ext4 noatime,users,exec 0 4 LABEL=root / ext4 noatime 0 1 LABEL=boot /boot ext4 noatime,rw,users 1 2 LABEL=home /home ext4 noatime,users,exec 0 4 UUID=3e54d3ee-ac61-4bce-ae90-1da94ec8bd78 none swap sw 0 0 LABEL=sda-portage /usr/portage ext4 noatime,users 0 3
I found on trying to build and compile a new kernel I was getting error messages from objdump along the lines of…
cd /usr/src/linux make oldconfig make -j9 ... objdump: arch/x86/kernel/acpi/.tmp_sleep.o: unable to initialize decompress status for section .debug_info objdump: arch/x86/kernel/acpi/.tmp_sleep.o: unable to initialize decompress status for section .debug_info objdump: arch/x86/kernel/acpi/.tmp_sleep.o: file format not recognized
This turned out to be because of an issue with a recent update to dev-libs/elfutils-0.175 and reverting to dev-libs/elfutils-0.173 solved the problem.