On Wed, 2020-11-04 at 14:05 -0800, John Jason Jordan wrote: > On Wed, 4 Nov 2020 13:18:50 -0800 > Russell Senior <[email protected]> dijo: > > > My nvme foo is weak, but I think your partition on an nvme is going > > to > > look like /dev/nvme1n1p1 or something. I just checked one of my > > machines and confirmed that naming. > > > > so, modify your mdadm command to use /dev/nvme1n1p1 /dev/nvme2n1p1 > > /dev/nvme3n1p1 /dev/nvme4n1p1 at the end. > > OK, you put your finger on one thing that bothered me about my > command > - how to name the devices. To figure this out better I checked the > net > for NVMe device naming, and found a lot of instructions. But first I > need to point out that at the moment all four drives have no > partitions. I assume that means that the 'p1' at the end would not be > used. > > As for the 'n1,' apparently this refers to a namespace, but all the > writers of stuff that I read about this made a serious error: They > assumed that all their readers knew what a namespace is. > Unfortunately, > yo no, pas moi, not me. > > sudo lsblk or sudo fdisk -l will tell you what names to use.
It is the same as with the /dev/sda-z if you created partitions because you want to use them - then use them. If you want to use full disk - either do not use partitions or make them cover full disk. As about advantages or disadvantages of btrfs/zfs - if you are decided on raid0 and you do not care about data safety (that comes with raid0 probably) - then there are no advantages to mdadm. btrfs and zfs are more flexible in some ways over mdadm - not really for basic JBOD/raid0/raid1/raid10 scenarios. btrfs is true libre/GNU and more flexible (you can remove disks) than zfs. For this reason Ubuntu is the only linux distro doing zfs out of the box. I hope that Canonical has some bullet proof deal with Oracle otherwise - see java vs. android forever court battle for damages greater than the value of the universe. All linux distros support btrfs. Otherwise btrfs/zfs difference is a wash. Zealots of either kind will disagree. In terms of check data/metadata check sums - mdadm/btrfs/zfs - it depends - you can format mdadm raid with btrfs and be fully covered with check sums (Synology does that currently). With zfs - you just go all the way with zfs, I suppose. If you have Synology NAS for backup - and use btrfs on it (new setups (last 3-4 years) probably do) - than it would make sense to use btrfs as you could do block level incremental btrfs snapshot replication. If you care. Other than that - I am not going to waste my time writing what you can read about elsewhere: https://duckduckgo.com/?t=ffsb&q=btrfs&ia=web https://duckduckgo.com/?q=zfs&t=ffsb&ia=web Tomas _______________________________________________ PLUG: https://pdxlinux.org PLUG mailing list [email protected] http://lists.pdxlinux.org/mailman/listinfo/plug
