df command output shows partly dm-X devices and partly lvm mapper devices,

shadow-fmx · 06-30-2015, 03:22 AM

Hi,

Actually, when I do an umount -l (must be lazy because stuff in use) and then a remount as configged in fstab
it shows up correct.
However after a reboot the correction is gone again.

Code:

root@debian64[pts/0]:/ # umount /usr -l                                                                                                                                                      10:19
root@debian64[pts/0]:/ # mount /dev/mapper/vg0-usr /usr                                                                                                                                      10:19
root@debian64[pts/0]:/ # df                                                                                                                                                                  10:20
Filesystem          1K-blocks    Used Available Use Mounted on
/dev/dm-0              943128  400044    477960  46% /
udev                    10240       0     10240   0% /dev
tmpfs                  411740   10888    400852   3% /run
tmpfs                 1029344       0   1029344   0% /dev/shm
tmpfs                    5120       0      5120   0% /run/lock
tmpfs                 1029344       0   1029344   0% /sys/fs/cgroup
/dev/mapper/vg0-srv  43121152   49032  40858644   1% /srv
/dev/mapper/vg0-var   1886280  711312   1061100  41% /var
/dev/mapper/vg0-tmp    943128    1228    876776   1% /tmp
tmpfs                  205872       0    205872   0% /run/user/0
/dev/mapper/vg0-usr   3776568 1101664   2463348  31% /usr

voleg · 06-30-2015, 03:36 AM

As I see, the "problem" is with devices mounted within initrd image.
No problem with these names at all. You should revise your boot image to fix names.
I am not sure that you will success, but you will learn a lot.

EDDY1 · 06-30-2015, 03:50 AM

not sure that separate /usr is a problem with lvm but with systemd /usr is not being used but you have 2 options rebuild initrd or initramfs (not sure which) to support separate /usr or move it to root.

BTW Thanks for posting this thread the machine that I'm on is having problems & it's most likely due to separate /usr.
Anything that I want to use I have to execute from /usr/bin

shadow-fmx · 06-30-2015, 04:02 AM

@voleg,

I actually tried unpacking/decompressing the image last weekend. It looked succesfull, but where to go from there .... ?

@eddy1

"rebuild initrd to support separate /usr".
I am not familliar with this process. Can you point me in the right direction ? (I can google it myself, but about 10 million links show up).
What is the core functionality of systemd ?

EDDY1 · 06-30-2015, 05:09 PM

That's my problem also. Still searching but threads about archlinux is probably where you would find your answer.
You may even find it in LFS book.

jnlq · 10-29-2015, 07:28 AM

I just did a release upgrade on my ubuntu box to 15.10 and now have the same issue... / is showing in df as /dev/dm-2

previously it showed as the /dev/mapper/raid1-root

my two other lv's are showing in df properly....

Filesystem Size Used Avail Use% Mounted on
udev 16G 4.0K 16G 1% /dev
tmpfs 3.2G 2.6M 3.2G 1% /run
/dev/dm-2 103G 9.8G 88G 10% /
none 4.0K 0 4.0K 0% /sys/fs/cgroup
none 5.0M 8.0K 5.0M 1% /run/lock
tmpfs 16G 0 16G 0% /run/shm
cgmfs 100K 0 100K 0% /run/cgmanager/fs
none 100M 0 100M 0% /run/user
/dev/mapper/raid5-home 443G 11G 414G 3% /home
/dev/mapper/raid5-data 5.0T 599G 4.2T 13% /data
/dev/sda1 228M 82M 134M 38% /boot
tmpfs 3.2G 0 3.2G 0% /run/user/1000
//nacho/Backups 5.5T 2.7T 2.9T 48% /mnt/nacho
//naive/data 8.2T 3.1T 5.1T 38% /mnt/naive
//nexus/data 11T 8.1T 2.8T 75% /mnt/nexus

Also, it looks like in dmesg that / is being mounted and then re-mounted as well.

Were any of you able to track this down and correct?

shadow-fmx · 10-29-2015, 07:34 AM

Not me.
I got lost in too much detail without knowing what exactly is causing this.
Still have this issue in my template server, either after updating or new (!) installation from iso.

jnlq · 10-29-2015, 11:30 AM

Very strange, as 'mount' shows them properly

but /proc/mounts is missing the mapper for / and showing the device instead...

from 'mount'

Code:

/dev/mapper/raid1-root on / type ext4 (rw,relatime,errors=remount-ro,data=ordered)

from '/proc/mounts' (and df)

Code:

/dev/dm-2 / ext4 rw,relatime,errors=remount-ro,data=ordered 0 0

and yet, both the other two lv's show fine in both...

Code:

/dev/mapper/raid5-home on /home type ext4 (rw,relatime,errors=remount-ro,data=ordered)
/dev/mapper/raid5-data on /data type ext4 (rw,relatime,errors=remount-ro,data=ordered)

Code:

/dev/mapper/raid5-home /home ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
/dev/mapper/raid5-data /data ext4 rw,relatime,errors=remount-ro,data=ordered 0 0

shadow-fmx · 10-29-2015, 11:38 AM

I just do not understand that only a few people seem to suffer from this.
If this would be some major bug I expected way more replies to this thread.

JeremyBoden · 10-29-2015, 11:53 AM

So presumably things like /var, /usr etc are unmounted?

If not, is it possible that you have two / directories, but you are not using one of them?

jnlq · 10-29-2015, 01:16 PM

Everything is fine and mounted on my box. I just cut the relative info out for the post.

my var, usr, etc, etc. are all on that lv (/)

I can't find anything on this elsewhere, not sure what to search for either though.

Oh, and other than this 'visual' issue, the system is fine. Just my monitoring scripts and such that look for the dm mapped named had to all be revised to point to the dev name instead - it's just bothering because I can't figure it out

shadow-fmx · 10-29-2015, 01:27 PM

I had the same feeling

WHY is this wrong all of a sudden....

JeremyBoden · 10-29-2015, 03:47 PM

So you have a Logical Volume that isn't mounted (supposedly)...

BTW My desktop, which has absolutely no LVM on it, at all, always gives

Code:

dmesg | grep 'EXT4-fs (sda2)'
[    2.767197] EXT4-fs (sda2): mounted filesystem with ordered data mode. Opts: (null)
[    6.765981] EXT4-fs (sda2): re-mounted. Opts: (null)
[    6.928737] EXT4-fs (sda2): re-mounted. Opts: errors=remount-ro

I read that as sda2 get mounted (rw), remounted (rw), remounted again (rw)...
sda2 is my / partition

Anyway a 'mount' command shows everything is good.

Code:

/dev/sda2 on / type ext4 (rw,relatime,errors=remount-ro,data=ordered)

Perhaps you have a similar "problem".

BTW How does the kernel know whether to mount / as (ro) or (rw) without first mounting it, so that it can see what the /etc/fstab file has to say about things?

jnlq · 10-29-2015, 04:42 PM

The LVM volume is mounted just fine. The /dev/mapper/LVNAME is a mapping to the long device name for ease of use. The dm-2 name is a very short name for the device itself, which isn't the proper method of displaying the LVM volume.

From what I understand the initial root volume is always mount RO at first then remounted with FSTAB options. So I believe that is functioning as designed.

More explanation on the /dev/mapper vs /dev/dm naming:

Well, from functional point of view they are same - they point to the
same device, of course. But dmsetup/lvm itself does not create the
/dev/dm-X nodes - the ones in /dev/mapper are the right and official
ones that should always be used.

The /dev/dm-X nodes are created by some general udev rules, dm-X is
only internal kernel name for that device and you can't rely on those
names (because the number X that is assigned is not stable and could
be changed - it depends on the sequence of device activation).

jnlq · 10-29-2015, 06:37 PM

maybe related? seems identical - going to try the patch tonight - will report back

https://bugs.debian.org/cgi-bin/bugr...cgi?bug=791754

EDIT: patch works! (may need to run update-initramfs -u -k all after - I did just to clean up all the other crud I tried)