Bug 1977734

It sounds like this is already fixed in 8.5 but I don't have nvme devices to test against to verify.  If you'd like to be sure it's fixed in 8.5 please try the most recent lvm build.

Hello,

Tried latest lvm2 build and it does not work.
reproducible with non nvme devices as well.

# parted /dev/mapper/mpathb mklabel gpt
Information: You may need to update /etc/fstab.

# pvcreate -y /dev/mapper/mpathb
  Cannot use /dev/mapper/mpathb: device is partitioned        <<<

# rpm -q lvm2
lvm2-2.03.12-2.el8.x86_64

# pvcreate --version
  LVM version:     2.03.12(2)-RHEL8 (2021-05-19)
  Library version: 1.02.177-RHEL8 (2021-05-19)
  Driver version:  4.43.0
  Configuration:   ./configure --build=x86_64-redhat-linux-gnu --host=x86_64-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --with-default-dm-run-dir=/run --with-default-run-dir=/run/lvm --with-default-pid-dir=/run --with-default-locking-dir=/run/lock/lvm --with-usrlibdir=/usr/lib64 --enable-fsadm --enable-write_install --with-user= --with-group= --with-device-uid=0 --with-device-gid=6 --with-device-mode=0660 --enable-pkgconfig --enable-cmdlib --enable-dmeventd --enable-blkid_wiping --with-cluster=internal --enable-cmirrord --with-udevdir=/usr/lib/udev/rules.d --enable-udev_sync --with-thin=internal --with-cache=internal --enable-lvmpolld --enable-lvmlockd-dlm --enable-lvmlockd-dlmcontrol --enable-lvmlockd-sanlock --enable-dbus-service --enable-notify-dbus --enable-dmfilemapd --with-writecache=internal --with-vdo=internal --with-vdo-format=/usr/bin/vdoformat --with-integrity=internal --disable-silent-rules



Thanks,
Rupesh

I've now learned what GPT labels/partitions are.

If a disk has any kind of partition header (GPT label alone, GPT label with partitions, or others), pvcreate will not overwrite that header, and reports:

# pvcreate /dev/sdc
  Cannot use /dev/sdc: device is partitioned

This seems correct to me, and it's consistent with how other tools handle the same device:

# blkid /dev/sdc
/dev/sdc: PTUUID="237eb497-c8b7-4e2b-b68f-c4ae47bad331" PTTYPE="gpt"

# mkfs.ext4 /dev/sdc
mke2fs 1.45.6 (20-Mar-2020)
Found a gpt partition table in /dev/sdc
Proceed anyway? (y,N)                 

# mkfs.xfs /dev/sdc
mkfs.xfs: /dev/sdc appears to contain a partition table (gpt).
mkfs.xfs: Use the -f option to force overwrite.


A different question is whether the pvcreate -f option or -ff option will allow a user to overwrite any partition header.  In the current version, pvcreate -f or -ff do not allow overwriting it.  This looks like a bug (and seems to be unrelated to either NVME or GPT.)

Here's a dev branch patch that allows pvcreate -f to overwrite partitions:
https://sourceware.org/git/?p=lvm2.git;a=shortlog;h=refs/heads/dev-dct-pvcreate-f

I was trying to track down a commit for this regression but haven't found that yet (got back to late 2019), to see if that reveals any other cases for -f that also need to be fixed.

I have gone back and tested 8.2, 8.1 and 8.0, and none of these versions allow pvcreate -f to overwrite a partition header.  

The stable-2.02 branch (basis for RHEL7 versions) also does not allow pvcreate -f to overwrite a partition header.

So, overwriting partitions with pvcreate -f sounds more like a feature request than a bug fix.

I suspect that the results shown above, where "pvcreate -f /dev/nvme0n1" overwrites a partition header, may in fact be a different issue (not recognizing nvme devices/partitions).  This patch for bug 1859659 fixed recognizing nvme devices/partitions:


commit bee9f4efdd8195f383f026290b741314cdc42439
Author: David Teigland <teigland>
Date:   Thu Dec 3 10:48:21 2020 -0600

    filter-mpath: work with nvme devices
    
    Recognize when a device is nvme, and apply filter-mpath to
    nvme devices in addition to scsi devices.

The debug trace shows that lvm does not see partitions on /dev/nvme0n1, so it is not excluded by filter-partitioned, which is meant to reject devices with partition tables.  This allows the device to be formatted, which includes calling the wiping code (which does find partitions to wipe and produces the messages.)

So, the problem is with the old code failing to see partition info on the device.  If partition detection was fixed in the old version, then the old version would disallow overwriting the partition tables.

Perhaps it makes the most sense to use this bz as an RFE to allow pvcreate -f to overwrite partition tables.

Changing this as RFE request.

Description of problem:

pvcreate -y should allows to wipe gpt partition label and initialize device as PV.

Actual results:

$ sudo parted /dev/nvme1n1 mklabel GPT
Information: You may need to update /etc/fstab.

$ sudo pvcreate -y /dev/nvme1n1             
  Cannot use /dev/nvme1n1: device is partitioned

Expected results:

pvcreate should wipe gpt label from disk as the device doesn't has any actual partition on it.

change is here:
https://sourceware.org/git/?p=lvm2.git;a=commit;h=d5a950ca67c106403054ecb847e226e8b5a7c30e

# grep sdc /proc/partitions
   8       32  976236544 sdc
   8       33    2097152 sdc1
   8       34  974138368 sdc2

# ls -l /dev/sdc*
brw-rw----. 1 root disk 8, 32 Jan 13 13:33 /dev/sdc
brw-rw----. 1 root disk 8, 33 Jan 13 13:33 /dev/sdc1
brw-rw----. 1 root disk 8, 34 Jan 13 13:33 /dev/sdc2

# pvcreate /dev/sdc
  Cannot use /dev/sdc: device is partitioned

# pvcreate -f /dev/sdc
  Wiping dos signature on /dev/sdc.
  Physical volume "/dev/sdc" successfully created.

# grep sdc /proc/partitions
   8       32  976236544 sdc

# ls -l /dev/sdc*
brw-rw----. 1 root disk 8, 32 Jan 13 13:34 /dev/sdc

# pvs /dev/sdc
  PV         VG Fmt  Attr PSize   PFree  
  /dev/sdc      lvm2 ---  931.01g 931.01g

> change is here:
> https://sourceware.org/git/?p=lvm2.git;a=commit;
> h=d5a950ca67c106403054ecb847e226e8b5a7c30e

That was the commit in the main branch, the commit in the stable branch for rhel8 is:
https://sourceware.org/git/?p=lvm2.git;a=commit;h=a5c37afdca97d6565ea02bc4bc7d52f360823cd3

*** Bug 2138459 has been marked as a duplicate of this bug. ***