#lvm — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #lvm, aggregated by home.social.
-
Декларативная разметка дисков в NixOS с помощью Disko: от конфига до готовой ФС
Каждый, кто хоть раз устанавливал Linux вручную, знает этот ритуал: загрузиться с Live‑образа, запустить fdisk или parted , создать таблицу разделов, отформатировать их в btrfs/ext4, создать субтома, а затем вручную примонтировать всё это в /mnt . Конечно, такой подход можно использовать и при установке NixOs. Но вопрос — зачем? Это не только полностью противоречит самой сути NixOS — системы, где всё должно быть описано декларативно в коде, но и существенно влияет вообще на удобство использования этой системы в будущем. Стоит вам захотеть развернуть точно такую же конфигурацию на другом сервере или ноутбуке, и весь ручной процесс разметки приходится повторять заново, вспоминая точные флаги монтирования вместо того, чтобы написать всего лишь одну команду. В этой статье мы этим вопросом и займёмся, а поможет нам в этом известная утилита disko. Мы напишем чистый декларативный конфигурационный файл в ДВУХ вариантах:
https://habr.com/ru/articles/1073844/
#nixos #disko #lvm #luks #разметка_диска #linux_для_новичков #декларативный_подход
-
When I started using Linux some 30 years ago, I used ext2 as its filesystem. That became ext3 and then ext4, and LVM came along. All the tools you need for basically everything you would ever need to do with filesystems. ReiserFS seemed promising, but... well... we all know the sad end of that.
I feel it's time to move to something more modern, just like the creator of ext2, ext3 and ext4 has done. There are a few candidates, but for me it boils down to ZFS or BTRFS.
I have a bit of experience with zfs, and once you master the basics, it's pretty powerful. But it's also hungry in RAM, and it's at its best at high loads.
...
View article View summary
When I started using Linux some 30 years ago, I used ext2 as its filesystem. That became ext3 and then ext4, and LVM came along. All the tools you need for basically everything you would ever need to do with filesystems. ReiserFS seemed promising, but... well... we all know the sad end of that.
I feel it's time to move to something more modern, just like the creator of ext2, ext3 and ext4 has done. There are a few candidates, but for me it boils down to ZFS or BTRFS.
I have a bit of experience with zfs, and once you master the basics, it's pretty powerful. But it's also hungry in RAM, and it's at its best at high loads.
ZFS vs BTRFS
Then there's btrfs, which seems to be really stable now. Does pretty much what zfs does, using a lot less RAM, but it only runs on Linux. That's what I use, but it would be nice to know that I can mount an fs on a different OS if I need to, say OpenBSD.
It's also not optimized for the high loads that zfs likes, which makes it much more appropriate for my small private environments. Points for btrfs.
Send-receive looks really nice. I haven't used it myself yet, but at work it's used a lot on zfs. Writable snapshots is a really nice thing to have, I can imagine using that for cloning dev-environments and running them on the actual original snapshot. That's btrfs-only, so more points for that.
A big booh for zfs is that it's not open enough to be in Debian, so you'll have to compile the kernel-modules every time you update your kernel. That takes more than enough time for me to get irritated.
So, guess it's time to get myself some experience with btrfs...
#^https://www.wundertech.net/btrfs-vs-zfs-comparison/
#^https://www.salvagedata.com/blog/btrfs-zfs-xfs-ext4-how-are-they-different
#btrfs #zfs #ext4 #LVM -
🧵⬆️ For example: The Arch wiki says¹ that when I give my home partition the very specific name `/dev/mapper/home` chosen by Poettering and not something arbitrary like `/dev/mapper/home_partition`, then systemd will create a symlink in `/dev/disk/by-designator/` and I can henceforth use the following names
- `/dev/disk/by-designator/home` to refer to the unlocked volume
- `/dev/disk/by-designator/home-luks` to refer to the encrypted partitionto make use of persistent naming methods and to become able to avoid having to hard-code UUIDs into `/etc/fstab`.
But this works only if I use the particular names that Poettering has in mind, like `home` (rather than something arbitrary that I come up with, like `davids_home` or `home_partition`). This is a constructed example to illustrate the point.
So far I can follow Poettering's thinking. But all of this falls apart as soon as LVM (Logical Volume Manager) comes into play.
I don't see the benefits of following Poettering's Discoverable Partitions Specification and thus don't see why I should follow it if I want to use LVM (Logical Volume Manager).
¹ https://wiki.archlinux.org/title/Dm-crypt/System_configuration#Using_systemd-cryptsetup-generator
-
#Linux Does someone understand how Lennart Poettering's Discoverable Partitions Specification¹ integrates with Logical Volume Manager (LVM)?
Poettering insists that I give my partitions very specific names so that systemd can auto-mount the partitions for me and `/etc/fstab` becomes obsolete, to name just one of the motivations/benefits. If I LUKS-encrypt my partitions, Poettering wants that I name my partitions either of these pre-defined names:²
- /dev/mapper/root
- /dev/mapper/usr
- /dev/mapper/home
- /dev/mapper/srv
- /dev/mapper/esp
- /dev/mapper/xbootldr
- /dev/mapper/swap
- /dev/mapper/root-verity
- /dev/mapper/root-verity-sig
- /dev/mapper/usr-verity
- /dev/mapper/usr-verity-sig
- /dev/mapper/tmp
- /dev/mapper/varI can follow Poettering's thinking (even if you or I don't need to agree with it) … so long as nobody ever uses LVM.
If I add a second SSD to my computer, create just a single partition on it, LUKS-encrypt it and add a volume group on top of LUKS, which I then split into, say, three logical volumes: `/home`, `/var`, and `/var/log` to construct just a random example.
- Why should I name the underlying device mapper file `/dev/mapper/home` when there's more than `/home` on that volume group?
- If I named the device mapper file something arbitrary like `/dev/mapper/davids-lvm-container`, then LVM also wouldn't work with Poettering's Discoverable Partitions Specification. If I use LVM for everything, then none of the benefits motivated by Poettering work, even if I do as Poettering suggests. And if none of the benefits works anyway, then why should I bother using the names that Poettering suggests rather than the names I find intuitive, even if that means I have to keep `/etc/fstab` around and keep doing the things manually that everyone's always been doing manually anway?My ability to follow Poettering's thinking falls apart when I pull LVM into the equation.
Does Poettering not use LVM (Logical Volume Manager)? Am I missing something?
If I could see any benefit in doing things the way Poettering suggests, I would do it. I just don't see the benefits (when using LVM at least).
¹ https://uapi-group.org/specifications/specs/discoverable_partitions_specification/
² https://man.archlinux.org/man/systemd.image-filter.7 -
#TIL (Today I Learned) that when you use LVM (Logical Volume Manager) on Linux (to abstract your storage), you should prefer the volume path
/dev/my_volume_group/my_logical_volume
provided by LVM over the more low-level volume path
/dev/mapper/my_volume_group-my_logical_volume
that is provided by the underlying device mapper framework.
You should never interact with the underlying device mapper framework directly, and rather interact indirectly with it through the LVM and/or cryptsetup abstractions because the filepaths provided by the device mapper framework are 'intended for "internal use" and subject to possible "change between releases and distributions",' as the Arch Wiki puts it.
Sources:
- https://man.archlinux.org/man/lvm.8#VALID_NAMES
- https://wiki.archlinux.org/title/LVM#LVM_building_blocks -
This video here https://www.youtube.com/watch?v=a1AXHpog9iI help me so much to udnerstand how to install #archlinux #LVM on #LUKS
I've been archlinux user since long time but I've nerver installed arch in this way, I'll do it in these next days in my new computer. -
What is, in your opinion, the best value for data alignment when creating a physical volume using LVM (Logical Volume Manager)?
pvcreate --dataalignment ?? /dev/mapper/whatever
What should ?? be in your opinion?
-
Teach me how to backup a layered volume
https://piefed.blahaj.zone/c/techsupport/p/875815/teach-me-how-to-backup-a-layered-volume
-
Note to self: Run "sync" before "systemctl reboot"...
Something just got killed and I have no idea what. Just hope it wasn't anything LVM cache and/or BTRFS related even though nothing else that accessed the volume was (supposed to be) running anymore.
-
I want a mariadb database to store a lot of census data in a SQL queriable way for members to do research with. And I want a container doing NFS over a btrfs raid 1 volume as a replacement for my NAS thats been running for a long time and might be best upgraded.
last time I used LVM was like... 2002 so, any #linux #lvm users wanna give advice on how to integrate these 2 spinning disks and 2 small ssds to get a nice useful variety of LVM based volumes? recommend any tutorials?
-
For all those people building #NAS boxes with huge CPUs and lots of #RAM because that's how Linus (of YouTube not #Linux) did it, I'm running a 50+ TiB NAS on a Celeron N5105 with 8GiB of RAM using #LVM and #XFS.
My highest loads are during backups (borg) when the system hits just 80% idle. Use your precious RAM for gaming.
-
Auch gut: 1TB einer Volume Group zuordnen aber dann nur 50GB davon nutzen.
Bei einer VM. Deren virtuelle Festplatten man jederzeit erweitern kann.
Vielleicht haben die einfach zu viel Storage über.
-
Der Kunde möchte eine "eigene Platte, die die Daten enthält" um die Nutzdaten einer Serveranwendung vom System zu trennen.
Gleichzeitig will er die "eigene Platte" in die LVM-Volume Group des Betriebssystems hinzufügen, um den Platz bequem auf die vorhandenen LVs verteilen zu können, falls nötig.
Eine LVM-Volume Group mit zwei virtuellen Festplatten, ohne Stripeset, ohne LVM-RAID, ohne Parität.
Damit das Backupprogramm bei der Wiederherstellung der großen Platte eines Tages Mist bauen kann und anschließend die komplette Volume Group defekt ist und die VM nicht mehr startet?
P.S.: Auf der primären Festplatte des Betriebssystems sind noch 125 weitere Partitionen möglich, also 125 potenzielle Erweiterungen der VG.
Muss man nicht verstehen... 🤔
-
My 100% LLM free guide to install Chimera Linux with KDE Plasma on top of a Snapper ready BTRFS file system that resides inside a Logical Vulume Manager Logical Volume inside a Volume Group inside a encrypted Linux Unified Key Setup (LUKS) partition is done.
You can read it on:
https://theonedoc.github.io/Chimera-Linux-LUKS-LVM-BTRFS/
If I missed something or a step is overcomplicated/suboptimal let me know.
Preferable not via github.
-
You know, maybe it would have been a smarter design to first luks encrypt each of the drives and then put btrfs directly on-top of that.
But then I'd loose out on the "conveniences" of #LVM. Like:
* esp. a "production ready" RAID5/6 (hopefully that'll become ready in btrfs one day too)
* and LVM cache, for SSD writethrough caching the entire thing.
* Less risk of screwing up the #LUKS encryption guarantees.
* LVM snapshotsBut speaking of it, I don't even know if #btrfs handles this better...
-
could it be that when start out with an #dmraid with 4 equal drives in #RAID 5 and then add 4 more similar drives at the same time) later that it'll basically RAID0 each of the drives in the RAID5?
(I didn't tell it anything other than lvextend -l 100%FREE, I think. It has been a while. Just wondered why the total capacity is missing two instead of one drive...)
Guess it's time to learn how to tell which topology #LVM choose internally from the lvs/pvs/vgs/... tools output.
-
Die ersten Tests gelten dem primären Ziel: weitere Redundanz an sicherem Ort.
Dazu wurde zuerst mal der #YubiKey initialisiert mit HMAC-SHA1 und sichergestellt, dass ich damit auch an mein #KeePassXC ran komme. ✅
Dann wurde der #FIDO2 Hash in dem #LUKS - verschlüsselten #LVM via #systemd abgelegt und testweise damit gebootet. ✅
Die Initialisierung von #TOTP muss ich aufschieben, aber diese Einträge aus KeePassXC zu klonen, hatte ich die Tage erst noch getestet. Das wird klappen.
Bis hierhin sieht es gut aus. Als nächstes kommt #NFC, das ist für mich Neuland...
-
Wyng Backup 🐦 v0.8.1 is out and available for download! https://codeberg.org/tasket/wyng-backup
Changes from v0.8.0 include bug fixes, faster data seeking over unallocated space, expanded archive repair capabilities and improved error handling.
Development is also progressing toward v0.9, with alpha2 now available for testing. Alpha versions are now named 'wyng09a' so they aren't accidentally run instead of stable releases. These versions also require '--test-write-v3' option before writing to archives created by Wyng stable.
#wyng #backup #release #lvm #btrfs #linux #qubes -
Der Zugriff auf #LUKS / #LVM auf meinem primären Notebook klappt nun mit beiden Yubikeys auch beim Booten, aber das war unter #Fedora 44 schon etwas tricky und ist noch nicht perfekt:
root@j20:~# systemd-cryptenroll /dev/nvme0n1p3 SLOT TYPE 0 fido2 1 password 2 password 3 fido2 4 recoveryWas noch nicht optimal läuft, ist die Eingabe der Passphrase für den Security Token beim Booten, die zwar funktioniert, aber nicht von der normalen Eingabe zu unterscheiden ist. Erst danach wird man zum Berühren des Tokens aufgefordert.
Kein Schimmer, wie man das noch verbessert. Aber das ist Jammern auf hohem Niveau...
-
Curious how common/popular it is to use #LVM partitions instead of standard #Partitions these days. I was just looking at the Arch Wiki page for [LVM](https://wiki.archlinux.org/title/LVM) and I was surprised by all of the disadvantages they called out. I feel like I had assumed it was mostly all advantages. I guess looking at the list of disadvantages, the only one I'm wondering about is the "potentially worse performance".
On your laptops or desktops, do you typically use:
-
Sicherheit zwischen 0 und 1: Festplattenverschlüsselung erklärt
-
Производительность софт NVMe рейдов на основе mdadm, LVM и ZFS при использовании iSER и NVMe-oF (100G version) — Part 2
Продолжение первой части, смысл данной статьи сохраняется, а именно показать максимальную производительность, когда вопрос сохранности данных решается репликами или бэкапами. Цель - протестировать производительность трёх систем объединения физических устройств в одно логическое систем при использовании iSER и NVMe-oF. Так как статья получилась объёмная, она поделена на 2 части. В этой части - тесты с использованием HCIbench и iSER\NVMe-oF
-
Производительность софт NVMe рейдов на основе mdadm, LVM и ZFS при использовании iSER и NVMe-oF (100G version) — Part 1
Смысл данной статьи показать максимальную производительность, когда вопрос сохранности данных решается репликами или бэкапами. Цель - протестировать производительность трёх систем объединения физических устройств в одно логическое систем при использовании iSER и NVMe-oF. Так как статья получилась объёмная, она поделена на 2 части, в первой идут тесты на уровне локального устройства + вступление. Во второй уже тесты с использованием HCIbench и iSER\NVMe-oF
-
Русская рулетка дедиков: Собираем RAID 1 + LVM на живой системе без Rescue-режима
Ситуация до боли знакома каждому, кто регулярно арендует выделенные серверы: вы оплачиваете счет, заходите по SSH и видите, что ОС установлена на /dev/sda, а второй диск просто болтается пустым. Никакого RAID, никакой отказоустойчивости. Умрет первый диск — ваши данные исчезнут. Мы в SoftStore идем по пути полного контроля над процессом. В этой статье разбираем пошаговый протокол: как собрать программный RAID 1 и LVM прямо на живой, работающей операционной системе без использования Rescue-режима и переустановки. В качестве бонуса — практика по управлению квотами, снапшотами и замене дисков в Production.
https://habr.com/ru/articles/1029498/
#RAID #LVM #mdadm #Linux #Debian #выделенный_сервер #миграция #системный_администратор #fstab #администрирование
-
Ein lieber Mensch hat mir seinen recht aktuellen Desktop vermacht, zu einem Preis für den aktuell vermutlich grad mal das RAM zu bekommen wäre. Das System darf dann mein fast volljähriges mit Xeao 550 i7 und GTX 770 ablösen.
Nur: Damals gab es noch kein Secure Boot, TPM und GPT und ich verzweifle daran auf dem neumodischem Schnickschnack ein #Arch mit #LVM auf #LUKS mit einem funktionierdem bootloader zu versehen.
Ob sich so Opa gefühlt hat, als er den VHS-Rekorder nicht programmieren konnte? -
Okay @kubuntu, I'm impressed.
The stability of LTS (locked package versions in between releases), nears stock and unmodified KDE Plasma 6, releases every 6 months on the dot...
And all of this without forcing user to Snap packages!
I think I found my Wayland daily driver. If only Calamares didn't crash when trying to manually partition LVM... But that's a bug in Calamares since like 2016, so unlikely to be fixed.
#ubuntu #kubuntu #plasma #daily #linux #foss #kde #flatpak #snap #lvm
-
Unification de l'installeur : Le projet va désormais utiliser l'installeur de #LMDE (le "live-installer") pour toutes ses versions, remplaçant ainsi #Ubiquity. Ce changement permet de mieux supporter l' #OEM, le #SecureBoot et #LVM / #LUKS.
Arrivée de "Alfa" (Mint 23) : Une version de développement nommée provisoirement "Alfa" est déjà sur les rails.