[gentoo-user] Systemd and lvm

[gentoo-user] Systemd and lvm

[Robin Atwood]

[-- Attachment #1: Type: text/plain, Size: 1287 bytes --]

Having observed all the ranting, I thought I would try systemd on a laptop. It 
actually seems to work quite well and it is a lot faster. However I am having 
trouble getting my LVM partitions mounted. I installed the LVM service unit 
from the Gentoo Wiki but it never completes, timing-out on a job that mounts 
/var. The VG is actually created by an initramfs and when systemd dumps you 
out to the emergency shell you can use lvs to see the volumes, /dev/mapper has 
all the correct devices and "dmsetup ls" shows the LVs. In fact, everything 
appears as it should, the partitions just don't get mounted. I circumvented 
this by putting "mount -a" in the lvm.service unit, which then completes and 
the mount jobs time-out. Everything seems to be OK but it is a bit of a 
kludge. One thing I notice is:

# udevadm info -p /dev/mapper/vg00-rootfs -q all
syspath not found

Udev seems not to know about the LVs. Any ideas?

TIA
-Robin
-- 
----------------------------------------------------------------------
Robin Atwood.

"Ship me somewheres east of Suez, where the best is like the worst,
 Where there ain't no Ten Commandments an' a man can raise a thirst"
         from "Mandalay" by Rudyard Kipling
----------------------------------------------------------------------









[-- Attachment #2: Type: text/html, Size: 5502 bytes --]

Re: [gentoo-user] Systemd and lvm

[Canek Peláez Valdés]

On Fri, Jan 4, 2013 at 10:11 AM, Robin Atwood
<robin.atwood@attglobal.net> wrote:
> Having observed all the ranting, I thought I would try systemd on a laptop.
> It actually seems to work quite well and it is a lot faster. However I am
> having trouble getting my LVM partitions mounted. I installed the LVM
> service unit from the Gentoo Wiki but it never completes, timing-out on a
> job that mounts /var. The VG is actually created by an initramfs and when
> systemd dumps you out to the emergency shell you can use lvs to see the
> volumes, /dev/mapper has all the correct devices and "dmsetup ls" shows the
> LVs. In fact, everything appears as it should, the partitions just don't get
> mounted. I circumvented this by putting "mount -a" in the lvm.service unit,
> which then completes and the mount jobs time-out. Everything seems to be OK
> but it is a bit of a kludge. One thing I notice is:
>
>
>
> # udevadm info -p /dev/mapper/vg00-rootfs -q all
>
> syspath not found
>
>
>
> Udev seems not to know about the LVs. Any ideas?

How did you create your initramfs? Have you tried dracut, with
DRACUT_MODULES="lvm"?

Regards.
-- 
Canek Peláez Valdés
Posgrado en Ciencia e Ingeniería de la Computación
Universidad Nacional Autónoma de México

Re: [gentoo-user] Systemd and lvm

[Robin Atwood]

On Friday 04 Jan 2013, Canek Peláez Valdés wrote:
> On Fri, Jan 4, 2013 at 10:11 AM, Robin Atwood
> 
> <robin.atwood@attglobal.net> wrote:
> > Having observed all the ranting, I thought I would try systemd on a
> > laptop. It actually seems to work quite well and it is a lot faster.
> > However I am having trouble getting my LVM partitions mounted. I
> > installed the LVM service unit from the Gentoo Wiki but it never
> > completes, timing-out on a job that mounts /var. The VG is actually
> > created by an initramfs and when systemd dumps you out to the emergency
> > shell you can use lvs to see the volumes, /dev/mapper has all the
> > correct devices and "dmsetup ls" shows the LVs. In fact, everything
> > appears as it should, the partitions just don't get mounted. I
> > circumvented this by putting "mount -a" in the lvm.service unit, which
> > then completes and the mount jobs time-out. Everything seems to be OK
> > but it is a bit of a kludge. One thing I notice is:
> > 
> > 
> > 
> > # udevadm info -p /dev/mapper/vg00-rootfs -q all
> > 
> > syspath not found
> > 
> > 
> > 
> > Udev seems not to know about the LVs. Any ideas?
> 
> How did you create your initramfs? Have you tried dracut, with
> DRACUT_MODULES="lvm"?
> 
> Regards.

I always use genkernel with LVM=YES in genkernel.conf. There is a thread about 
the udev issue at http://forums.gentoo.org/viewtopic-p-6837888.html . I tried 
the suggested work-around but it made no difference, I must still use "mount -
a".

-Robin
-- 
----------------------------------------------------------------------
Robin Atwood.

"Ship me somewheres east of Suez, where the best is like the worst,
 Where there ain't no Ten Commandments an' a man can raise a thirst"
         from "Mandalay" by Rudyard Kipling
----------------------------------------------------------------------

Re: [gentoo-user] Systemd and lvm

[Canek Peláez Valdés]

On Sat, Jan 5, 2013 at 7:36 AM, Robin Atwood <robin.atwood@attglobal.net> wrote:
> On Friday 04 Jan 2013, Canek Peláez Valdés wrote:
>> On Fri, Jan 4, 2013 at 10:11 AM, Robin Atwood
>>
>> <robin.atwood@attglobal.net> wrote:
>> > Having observed all the ranting, I thought I would try systemd on a
>> > laptop. It actually seems to work quite well and it is a lot faster.
>> > However I am having trouble getting my LVM partitions mounted. I
>> > installed the LVM service unit from the Gentoo Wiki but it never
>> > completes, timing-out on a job that mounts /var. The VG is actually
>> > created by an initramfs and when systemd dumps you out to the emergency
>> > shell you can use lvs to see the volumes, /dev/mapper has all the
>> > correct devices and "dmsetup ls" shows the LVs. In fact, everything
>> > appears as it should, the partitions just don't get mounted. I
>> > circumvented this by putting "mount -a" in the lvm.service unit, which
>> > then completes and the mount jobs time-out. Everything seems to be OK
>> > but it is a bit of a kludge. One thing I notice is:
>> >
>> >
>> >
>> > # udevadm info -p /dev/mapper/vg00-rootfs -q all
>> >
>> > syspath not found
>> >
>> >
>> >
>> > Udev seems not to know about the LVs. Any ideas?
>>
>> How did you create your initramfs? Have you tried dracut, with
>> DRACUT_MODULES="lvm"?
>>
>> Regards.
>
> I always use genkernel with LVM=YES in genkernel.conf. There is a thread about
> the udev issue at http://forums.gentoo.org/viewtopic-p-6837888.html . I tried
> the suggested work-around but it made no difference, I must still use "mount -
> a".

I've never used genkernel. You could try dracut; its mandatory
dependencies are minimal, and it's actually designed to create an
initramfs, not like genkernel, where the functionality was added as an
afterthought.

Another option is to roll your own initramfs, like the first responder
in the forums thread.

Good luck.
-- 
Canek Peláez Valdés
Posgrado en Ciencia e Ingeniería de la Computación
Universidad Nacional Autónoma de México

Re: [gentoo-user] Systemd and lvm

[Robin Atwood]

[-- Attachment #1: Type: text/plain, Size: 2593 bytes --]

On Sunday 06 January 2013, Canek Peláez Valdés wrote:
> On Sat, Jan 5, 2013 at 7:36 AM, Robin Atwood <robin.atwood@attglobal.net> 
wrote:
> > On Friday 04 Jan 2013, Canek Peláez Valdés wrote:
> >> On Fri, Jan 4, 2013 at 10:11 AM, Robin Atwood
> >> 
> >> <robin.atwood@attglobal.net> wrote:
> >> > Having observed all the ranting, I thought I would try systemd on a
> >> > laptop. It actually seems to work quite well and it is a lot faster.
> >> > However I am having trouble getting my LVM partitions mounted. I
> >> > installed the LVM service unit from the Gentoo Wiki but it never
> >> > completes, timing-out on a job that mounts /var. The VG is actually
> >> > created by an initramfs and when systemd dumps you out to the
> >> > emergency shell you can use lvs to see the volumes, /dev/mapper has
> >> > all the correct devices and "dmsetup ls" shows the LVs. In fact,
> >> > everything appears as it should, the partitions just don't get
> >> > mounted. I circumvented this by putting "mount -a" in the lvm.service
> >> > unit, which then completes and the mount jobs time-out. Everything
> >> > seems to be OK but it is a bit of a kludge. One thing I notice is:
> >> > 
> >> > 
> >> > 
> >> > # udevadm info -p /dev/mapper/vg00-rootfs -q all
> >> > 
> >> > syspath not found
> >> > 
> >> > 
> >> > 
> >> > Udev seems not to know about the LVs. Any ideas?
> >> 
> >> How did you create your initramfs? Have you tried dracut, with
> >> DRACUT_MODULES="lvm"?
> >> 
> >> Regards.
> > 
> > I always use genkernel with LVM=YES in genkernel.conf. There is a thread
> > about the udev issue at
> > http://forums.gentoo.org/viewtopic-p-6837888.html . I tried the
> > suggested work-around but it made no difference, I must still use "mount
> > - a".
> 
> I've never used genkernel. You could try dracut; its mandatory
> dependencies are minimal, and it's actually designed to create an
> initramfs, not like genkernel, where the functionality was added as an
> afterthought.
> 
> Another option is to roll your own initramfs, like the first responder
> in the forums thread.
> 
> Good luck.

Maybe I will try dracut but I suspect the problem lies with systemd.

-Robin
-- 
----------------------------------------------------------------------
Robin Atwood.

"Ship me somewheres east of Suez, where the best is like the worst,
 Where there ain't no Ten Commandments an' a man can raise a thirst"
         from "Mandalay" by Rudyard Kipling
----------------------------------------------------------------------









[-- Attachment #2: Type: text/html, Size: 13453 bytes --]

[gentoo-user] systemd and lvm

[Stefan G. Weichinger]

systemd-204
lvm2-2.0.2.99-r2

(lvm2 patched as mentioned in

https://bugs.gentoo.org/show_bug.cgi?id=480066

this should only matter for activation)

-

See what happens here:

# lvcreate -n mlp-tmpl -L 11G VG02
  /dev/VG02/mlp-tmpl: not found: device not cleared
  Aborting. Failed to wipe start of new LV.
  device-mapper: remove ioctl on  failed: Das Gerät oder die Ressource
ist belegt
  semid 163841: semop failed for cookie 0xd4d92b7: incorrect semaphore state
  Failed to set a proper state for notification semaphore identified by
cookie value 223187639 (0xd4d92b7) to initialize waiting for incoming
notifications.


# lvcreate -n mlp-tmpl -L 11G -Zn VG02
  WARNING: "mlp-tmpl" not zeroed
  Logical volume "mlp-tmpl" created
  semid 196609: semop failed for cookie 0xd4d74cd: incorrect semaphore state
  Failed to set a proper state for notification semaphore identified by
cookie value 223179981 (0xd4d74cd) to initialize waiting for incoming
notifications.


https://bugzilla.redhat.com/show_bug.cgi?id=727925

talks of udev_sync=1 ... checked that, is there ...

CONFIG_NET_NS=y (whatever that is ... do I need it?)

...

Does anyone know what to do? I wanted to ask here before I file a bug at
bgo ...

Would be nice to solve this before I set up my shiny new KVM host where
I heavily want to use LVM-snapshots for backups. Sure, I could also stay
with openrc, correct!

(didn't mention: had that semaphore stuff on my main workstation as
well, only when booting via systemd)

Thanks, Stefan

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
> Failed to set a proper state for notification semaphore identified by
> cookie value


Also found this:

http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f845f96ca4cf1a9c17d73501e232d

I have

# zgrep UEV /proc/config.gz
CONFIG_UEVENT_HELPER_PATH=""
CONFIG_DM_UEVENT=y

so this is not my solution here ...

Re: [gentoo-user] systemd and lvm

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 601 bytes --]

On Wednesday 11 Sep 2013 12:38:23 Stefan G. Weichinger wrote:
> Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
> > Failed to set a proper state for notification semaphore identified by
> > cookie value
> 
> Also found this:
> 
> http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f845f
> 96ca4cf1a9c17d73501e232d
> 
> I have
> 
> # zgrep UEV /proc/config.gz
> CONFIG_UEVENT_HELPER_PATH=""
> CONFIG_DM_UEVENT=y
> 
> so this is not my solution here ...

I wonder if adding CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" would help you 
here.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 12.09.2013 08:50, schrieb Mick:
> On Wednesday 11 Sep 2013 12:38:23 Stefan G. Weichinger wrote:
>> Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
>>> Failed to set a proper state for notification semaphore
>>> identified by cookie value
>> 
>> Also found this:
>> 
>> http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f845f
>>
>> 
96ca4cf1a9c17d73501e232d
>> 
>> I have
>> 
>> # zgrep UEV /proc/config.gz CONFIG_UEVENT_HELPER_PATH="" 
>> CONFIG_DM_UEVENT=y
>> 
>> so this is not my solution here ...
> 
> I wonder if adding CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" would
> help you here.

I don't have that binary.  And some page on my way said the contrary:
set it to empty and let udev (?) do that.

Re: [gentoo-user] systemd and lvm

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 1119 bytes --]

On Thursday 12 Sep 2013 09:37:32 Stefan G. Weichinger wrote:
> Am 12.09.2013 08:50, schrieb Mick:
> > On Wednesday 11 Sep 2013 12:38:23 Stefan G. Weichinger wrote:
> >> Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
> >>> Failed to set a proper state for notification semaphore
> >>> identified by cookie value
> >> 
> >> Also found this:
> >> 
> >> http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f8
> >> 45f
> 
> 96ca4cf1a9c17d73501e232d
> 
> >> I have
> >> 
> >> # zgrep UEV /proc/config.gz CONFIG_UEVENT_HELPER_PATH=""
> >> CONFIG_DM_UEVENT=y
> >> 
> >> so this is not my solution here ...
> > 
> > I wonder if adding CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" would
> > help you here.
> 
> I don't have that binary.  And some page on my way said the contrary:
> set it to empty and let udev (?) do that.

Ha!  Neither do I!

# ls -la /sbin/hotplug
ls: cannot access /sbin/hotplug: No such file or directory

I can honestly say that I can't remember filling in this entry when 
configuring my kernels, but then how did it get there?

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 12.09.2013 14:43, schrieb Mick:

>> I don't have that binary.  And some page on my way said the
>> contrary: set it to empty and let udev (?) do that.
> 
> Ha!  Neither do I!
> 
> # ls -la /sbin/hotplug ls: cannot access /sbin/hotplug: No such
> file or directory
> 
> I can honestly say that I can't remember filling in this entry when
>  configuring my kernels, but then how did it get there?
> 

dunno.

Kernel help says about that setting:

> Before the switch to the netlink-based uevent source, this was
> │ │ used to hook hotplug scripts into kernel device events. It
> │ │ usually pointed to a shell script at /sbin/hotplug.
> │ │ This should not be used today, because usual systems create
> │ │ many events at bootup or device discovery in a very short time
> │ │ frame. One forked process per event can create so many
> processes
> │ │ that it creates a high system load, or on smaller systems
> │ │ it is known to create out-of-memory situations during bootup.


I also found configs having this:

CONFIG_UEVENT_HELPER_PATH="/usr/bin/udevadm"

That binary would exist here.

I am unsure if I should try that.


S

Re: [gentoo-user] systemd and lvm

[Canek Peláez Valdés]

[-- Attachment #1: Type: text/plain, Size: 1389 bytes --]

On Sep 12, 2013 8:04 AM, "Stefan G. Weichinger" <lists@xunil.at> wrote:
>
> Am 12.09.2013 14:43, schrieb Mick:
>
> >> I don't have that binary.  And some page on my way said the
> >> contrary: set it to empty and let udev (?) do that.
> >
> > Ha!  Neither do I!
> >
> > # ls -la /sbin/hotplug ls: cannot access /sbin/hotplug: No such
> > file or directory
> >
> > I can honestly say that I can't remember filling in this entry when
> >  configuring my kernels, but then how did it get there?
> >
>
> dunno.
>
> Kernel help says about that setting:
>
> > Before the switch to the netlink-based uevent source, this was
> > │ │ used to hook hotplug scripts into kernel device events. It
> > │ │ usually pointed to a shell script at /sbin/hotplug.
> > │ │ This should not be used today, because usual systems create
> > │ │ many events at bootup or device discovery in a very short time
> > │ │ frame. One forked process per event can create so many
> > processes
> > │ │ that it creates a high system load, or on smaller systems
> > │ │ it is known to create out-of-memory situations during bootup.
>
>
> I also found configs having this:
>
> CONFIG_UEVENT_HELPER_PATH="/usr/bin/udevadm"
>
> That binary would exist here.
>
> I am unsure if I should try that.

Don't, that only will create potential fork bombs.

Regards.

[-- Attachment #2: Type: text/html, Size: 1808 bytes --]

Re: [gentoo-user] systemd and lvm

[gottlieb]

On Thu, Sep 12 2013, Stefan G. Weichinger wrote:

> Am 12.09.2013 08:50, schrieb Mick:
>> On Wednesday 11 Sep 2013 12:38:23 Stefan G. Weichinger wrote:
>>> Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
>>>> Failed to set a proper state for notification semaphore
>>>> identified by cookie value
>>> 
>>> Also found this:
>>> 
>>> http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f845f
>>>
>>> 
> 96ca4cf1a9c17d73501e232d
>>> 
>>> I have
>>> 
>>> # zgrep UEV /proc/config.gz CONFIG_UEVENT_HELPER_PATH="" 
>>> CONFIG_DM_UEVENT=y
>>> 
>>> so this is not my solution here ...
>> 
>> I wonder if adding CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" would
>> help you here.
>
> I don't have that binary.  And some page on my way said the contrary:
> set it to empty and let udev (?) do that.

The wiki says to have it blank.  I just started working with the
systemd-wiki people and this is unsettled.  Some are using
/sbin/hotplug.  I believe there is not much experience to go on.
I will be trying to go to systemd with /sbin/hotplug.

hth,
allan

Re: [gentoo-user] systemd and lvm

[Canek Peláez Valdés]

On Thu, Sep 12, 2013 at 11:10 AM,  <gottlieb@nyu.edu> wrote:
> On Thu, Sep 12 2013, Stefan G. Weichinger wrote:
>
>> Am 12.09.2013 08:50, schrieb Mick:
>>> On Wednesday 11 Sep 2013 12:38:23 Stefan G. Weichinger wrote:
>>>> Am 11.09.2013 13:22, schrieb Stefan G. Weichinger:
>>>>> Failed to set a proper state for notification semaphore
>>>>> identified by cookie value
>>>>
>>>> Also found this:
>>>>
>>>> http://forums.gentoo.org/viewtopic-t-965446-view-previous.html?sid=5c1f845f
>>>>
>>>>
>> 96ca4cf1a9c17d73501e232d
>>>>
>>>> I have
>>>>
>>>> # zgrep UEV /proc/config.gz CONFIG_UEVENT_HELPER_PATH=""
>>>> CONFIG_DM_UEVENT=y
>>>>
>>>> so this is not my solution here ...
>>>
>>> I wonder if adding CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug" would
>>> help you here.
>>
>> I don't have that binary.  And some page on my way said the contrary:
>> set it to empty and let udev (?) do that.
>
> The wiki says to have it blank.  I just started working with the
> systemd-wiki people and this is unsettled.  Some are using
> /sbin/hotplug.  I believe there is not much experience to go on.
> I will be trying to go to systemd with /sbin/hotplug.

Do you even have /sbin/hotplug? I don't, in any of my machines, and I
don't even remember when was the last time I saw it.

From the git live systemd README [1]:

Legacy hotplug slows down the system and confuses udev:
          CONFIG_UEVENT_HELPER_PATH=""

From the kernel own help file:

config UEVENT_HELPER_PATH
        string "path to uevent helper"
        default ""
        help
         Path to uevent helper program forked by the kernel for
          every uevent.

          Before the switch to the netlink-based uevent source, this was
          used to hook hotplug scripts into kernel device events. It
          usually pointed to a shell script at /sbin/hotplug.
          This should not be used today, because usual systems create
          many events at bootup or device discovery in a very short time
          frame. One forked process per event can create so many processes
          that it creates a high system load, or on smaller systems
          it is known to create out-of-memory situations during bootup.

          To disable user space helper program execution at early boot
          time specify an empty string here. This setting can be altered
          via /proc/sys/kernel/hotplug or via /sys/kernel/uevent_helper
          later at runtime.

Really, whomever is recommending to set CONFIG_UEVENT_HELPER_PATH is
probably wrong. I can't find *one* place where it is recommended, and
several where they explicitly say to leave the option in blank.

Regards.

[1] http://cgit.freedesktop.org/systemd/systemd/tree/README
-- 
Canek Peláez Valdés
Posgrado en Ciencia e Ingeniería de la Computación
Universidad Nacional Autónoma de México

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 12.09.2013 18:22, schrieb Canek Peláez Valdés:

> Really, whomever is recommending to set CONFIG_UEVENT_HELPER_PATH is
> probably wrong. I can't find *one* place where it is recommended, and
> several where they explicitly say to leave the option in blank.

So ... I agree with this.

What to do about the initial problem then?

With openrc lvcreate is no problem, with systemd it is ... (for me, on 2
machines).

Stefan

Re: [gentoo-user] systemd and lvm

[gottlieb]

On Thu, Sep 12 2013, Canek Peláez Valdés wrote:

>
> Really, whomever is recommending to set CONFIG_UEVENT_HELPER_PATH is
> probably wrong. I can't find *one* place where it is recommended, and
> several where they explicitly say to leave the option in blank.

OK.  The wiki will continue to say it should be blank.
allan

Re: [gentoo-user] systemd and lvm

[Canek Peláez Valdés]

On Thu, Sep 12, 2013 at 12:35 PM, Stefan G. Weichinger <lists@xunil.at> wrote:
> Am 12.09.2013 18:22, schrieb Canek Peláez Valdés:
>
>> Really, whomever is recommending to set CONFIG_UEVENT_HELPER_PATH is
>> probably wrong. I can't find *one* place where it is recommended, and
>> several where they explicitly say to leave the option in blank.
>
> So ... I agree with this.
>
> What to do about the initial problem then?
>
> With openrc lvcreate is no problem, with systemd it is ... (for me, on 2
> machines).

Stefan, what initramfs are you using?

I don't have much experience with LVM; I just installed a Qemu virtual
machine with it, and it gave me no problems. But it was a dead simple
setup.

Your setup, however, seems to be rather complicated: you have LVM,
LUKS and (if I remember correctly) software RAID?

In my virtual machine I didn't had to do anything. There are no
services for LVM, and there are no scripts doing nothing LVM related.
There are a couple of udev rules, which I never touched, and systemd
together with that seems to handle everything by itself. I have
everything in LVM, including / and /boot (which is inside /).

Could you please explain how is exactly your layout? From drives to
partitions to PVs, VGs and LVs? And throw in there also the LUKS and
RAID (if used) setup. I will try to replicate that in a VM. Next week,
since we have a holiday weekend coming.

When I installed my LVM setup, I was surprised to find how easy it
was. The only problem I got was to install GRUB2 in /dev/vda, and even
that wasn't that difficult, and only qemu related. So perhaps the
problem is *moving* LVM machines to systemd, with the cruft from
previous OpenRC installations. Perhaps you don't need to *add*
anything, but to *remove* things that are not necessary anymore since
systemd+dracut handles everything.

Please explain to me your drive layout, so I can try to replicate it.

Regards.
-- 
Canek Peláez Valdés
Posgrado en Ciencia e Ingeniería de la Computación
Universidad Nacional Autónoma de México

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 12.09.2013 20:23, schrieb Canek Peláez Valdés:

> Stefan, what initramfs are you using?

dracut, run via your kerninst-script.

> Could you please explain how is exactly your layout? From drives to
> partitions to PVs, VGs and LVs? And throw in there also the LUKS and
> RAID (if used) setup. I will try to replicate that in a VM. Next week,
> since we have a holiday weekend coming.

thanks for your offer.

I wil happily list my setup BUT let me tell at first that the latest
sys-fs/lvm2-2.02.100 seems to have fixed that semaphore-issue.

After booting my desktop with it I quickly tested:

# lvcreate -n test -L1G VG03
  Logical volume "test" created
#

fine!

Three times ok ...

But I still face the fact that the LVs weren't activated at boot time.
Manual "vgchange -ay" needed ... or that self-written lvm.service
enabled as mentioned somewhat earlier.

Here my setup:

# lsblk
NAME                 MAJ:MIN RM   SIZE RO TYPE  MOUNTPOINT
sda                    8:0    0 931,5G  0 disk
├─sda1                 8:1    0     2M  0 part
├─sda2                 8:2    0     2G  0 part  [SWAP]
├─sda3                 8:3    0   600G  0 part
│ └─md127              9:127  0 595,1G  0 raid1
│   ├─VG03-music     253:0    0   190G  0 lvm   /mnt/music
│   ├─VG03-platz     253:1    0   200G  0 lvm   /mnt/platz
│   ├─VG03-media     253:2    0    45G  0 lvm   /mnt/media
│   ├─VG03-home      253:3    0    30G  0 lvm
│   ├─VG03-oopsfiles 253:4    0    12G  0 lvm   /mnt/oopsfiles
│   ├─VG03-dropbox   253:5    0     5G  0 lvm   /mnt/dropbox
│   ├─VG03-distfiles 253:6    0    10G  0 lvm   /usr/portage/distfiles
│   ├─VG03-gentoo32  253:7    0    15G  0 lvm   /mnt/gentoo32
│   ├─VG03-xp        253:8    0    40G  0 lvm
│   └─VG03-test      253:9    0     1G  0 lvm
└─sda6                 8:6    0    50G  0 part
  └─md4                9:4    0    50G  0 raid1
sdb                    8:16   0 931,5G  0 disk
├─sdb1                 8:17   0   100M  0 part
├─sdb2                 8:18   0  98,8G  0 part
├─sdb3                 8:19   0    50G  0 part
│ └─md4                9:4    0    50G  0 raid1
├─sdb4                 8:20   0  12,4G  0 part
└─sdb6                 8:22   0 595,1G  0 part
  └─md127              9:127  0 595,1G  0 raid1
    ├─VG03-music     253:0    0   190G  0 lvm   /mnt/music
    ├─VG03-platz     253:1    0   200G  0 lvm   /mnt/platz
    ├─VG03-media     253:2    0    45G  0 lvm   /mnt/media
    ├─VG03-home      253:3    0    30G  0 lvm
    ├─VG03-oopsfiles 253:4    0    12G  0 lvm   /mnt/oopsfiles
    ├─VG03-dropbox   253:5    0     5G  0 lvm   /mnt/dropbox
    ├─VG03-distfiles 253:6    0    10G  0 lvm   /usr/portage/distfiles
    ├─VG03-gentoo32  253:7    0    15G  0 lvm   /mnt/gentoo32
    ├─VG03-xp        253:8    0    40G  0 lvm
    └─VG03-test      253:9    0     1G  0 lvm
sdc                    8:32   0  55,9G  0 disk
├─sdc1                 8:33   0    25G  0 part  /
├─sdc2                 8:34   0     2G  0 part
└─sdc3                 8:35   0  28,9G  0 part  /home
sr0                   11:0    1  1024M  0 rom



This pretty much says it all, right?

2 hdds sda and sdb
1 ssd sdc

root-fs and /home on ssd ...

sda and sdb build two RAID-arrays (rather ugly names and partitions ...
grown over time):

# cat /proc/mdstat
Personalities : [raid1]
md4 : active raid1 sdb3[0] sda6[2]
      52395904 blocks super 1.2 [2/2] [UU]

md127 : active raid1 sdb6[0] sda3[1]
      623963072 blocks [2/2] [UU]

unused devices: <none>


# pvs
  PV         VG   Fmt  Attr PSize   PFree
  /dev/md127 VG03 lvm2 a--  595,05g 47,05g


--- thanks in advance, Stefan

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

new info here (for me):

https://bugs.gentoo.org/show_bug.cgi?id=480066#c19

gotta test ... right now I don't have the time.

S

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 13.09.2013 14:54, schrieb Stefan G. Weichinger:
> 
> new info here (for me):
> 
> https://bugs.gentoo.org/show_bug.cgi?id=480066#c19
> 
> gotta test ... right now I don't have the time.

first tests with "genkernel --udev ..." : negative.

More details maybe later this evening.

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 13.09.2013 15:33, schrieb Stefan G. Weichinger:
> Am 13.09.2013 14:54, schrieb Stefan G. Weichinger:
>>
>> new info here (for me):
>>
>> https://bugs.gentoo.org/show_bug.cgi?id=480066#c19
>>
>> gotta test ... right now I don't have the time.
> 
> first tests with "genkernel --udev ..." : negative.
> 
> More details maybe later this evening.

I found something.

journalctl -b told me that systemd was looking for

/usr/sbin/lvm

which does not exist.

I linked it from /sbin/lvm and this seems to help ... I still don't know
exactly where this comes from ... still digging.

I also removed lvm2 completely ... checked for lvm-related unit-files
and emerged it again ... no /usr/sbin/lvm found with grep. Maybe
compiled into ...

/usr/lib/systemd/system-generators/lvm2-activation-generator

Will have to check the sources or similar.

Re: [gentoo-user] systemd and lvm

[Stefan G. Weichinger]

Am 13.09.2013 19:36, schrieb Stefan G. Weichinger:

> /usr/sbin/lvm
> 
> which does not exist.
> 
> I linked it from /sbin/lvm and this seems to help ... I still don't know
> exactly where this comes from ... still digging.
> 
> I also removed lvm2 completely ... checked for lvm-related unit-files
> and emerged it again ... no /usr/sbin/lvm found with grep. Maybe
> compiled into ...
> 
> /usr/lib/systemd/system-generators/lvm2-activation-generator
> 
> Will have to check the sources or similar.

Oh ... much going on here:

https://bugs.gentoo.org/show_bug.cgi?id=484752