as newer kernels instruct udev to set the name of /dev/mapper/control device itself, udev auto-assigns NAME="mapper/control", so this rule from /etc/udev/rules.d/64-device-mapper.rules adds a symlink that would overwrite the device: KERNEL=="device-mapper", SYMLINK+="mapper/control" So udev prints a warning like this at boot: link_update: creating link '/dev/mapper/control' to '/dev/mapper/control' node_symlink: found existing node instead of symlink '/dev/mapper/control' node_symlink: device node '/dev/mapper/control' already exists, link to '/dev/mapper/control' will not overwrite it Easiest solution is to do the same as kernel developers did in kernel-space :change the rule to KERNEL=="device-mapper", NAME="mapper/control" Other versions are to strip of this rule here and add it to udev-provided compat-rules (same effect but different file the rule is in). But that creates new dependencies (and I like to avoid it). If lvm2 does require new enough udev (did not check) the file can also be moved to /lib/udev/rules.d, so we can some time in the future remove the CONFIG_PROTECT_MASK for /etc/udev/rules.d and open the dir up for just local admin usage.
agk: Can I get opinion on changing your upstream udev rules as needed per this bug?
(In reply to comment #1) > Can I get opinion on changing your upstream udev rules as needed per this bug? Well, we have KERNEL=="device-mapper", NAME="mapper/control" in the upstream...
For me, setting NAME="mapper/control" fixed the issue...
anyone going to jump out there and fix this?
Commited. Long term we can have a look at the rules upstream provides.
(In reply to comment #5) > Commited. May I ask you which udev version contained your commitment ? I updated the udev package some days ago to sys-fs/udev-149 and now I'm facing this message.
I can confirm this. After updating to version 149 I received the warning. Fixing the udev rules as indicated above fixed the problem. The question remains, why would a version be unmasked with an issue that has allegedly been dealt with.
(In reply to comment #6) > (In reply to comment #5) > > Commited. > May I ask you which udev version contained your commitment ? As you may guess from the subject line of this bug: The lvm-related udev-rules are contained in lvm2, and the bug is in sys-fs/lvm2-2.02.56-r1 and -r2 and is fixed in lvm2-2.02.56-r3: *lvm2-2.02.56-r3 (05 Jan 2010) 05 Jan 2010; Matthias Schwarzott <zzam@gentoo.org> +files/64-device-mapper.rules-2.02.56-r3, +lvm2-2.02.56-r3.ebuild: Fix udev rule for mapper/control device, Bug #295186. So I suggest lvm2-2.02.56-r1 could be stabled. Nevertheless this is just a warning, nothing critical, just poluting boot console with text.
(In reply to comment #8) > (In reply to comment #6) >> (In reply to comment #5) >>> Commited. >> May I ask you which udev version contained your commitment ? > As you may guess from the subject line of this bug: The lvm-related udev-rules > are contained in lvm2, and the bug is in sys-fs/lvm2-2.02.56-r1 and -r2 and is > fixed in lvm2-2.02.56-r3: > > *lvm2-2.02.56-r3 (05 Jan 2010) > > 05 Jan 2010; Matthias Schwarzott <zzam@gentoo.org> > +files/64-device-mapper.rules-2.02.56-r3, +lvm2-2.02.56-r3.ebuild: > Fix udev rule for mapper/control device, Bug #295186. > > So I suggest lvm2-2.02.56-r1 could be stabled. I assume you mean -r3 could be stabilized? I just ran into this, now 2 *months* later... > Nevertheless this is just a warning, nothing critical, just poluting boot > console with text. Hopefully you are right. I really dislike unmasking additional critical system packages to unstable just to fix stupid bugs that are already fixed, and I don't understand the recommended fixes above enough to do them manually.
hello, unfortunately, i do not agree status = resolved fixed :-( i have faced the issue longer then 1 year (have a look at my post http://forums.gentoo.org/viewtopic-t-842415-highlight-.html) and i am really quite nerves because the warning during boot are still here and there is the right (fixed) rule in /etc/udev/rules.d/64-device-mapper.rules: KERNEL=="device-mapper", NAME="mapper/control" i am using stable udev-151-r4 and lvm2-2.02.73-r1 on 2.6.38-gentoo-r6 #1 SMP Sat May 28 11:33:22 CEST 2011 x86_64 Intel(R) Core(TM)2 Quad CPU Q9450 @ 2.66GHz GenuineIntel GNU/Linux platform. today i have created a new lv using lvcreate -L8G -nusrsrc raid5 and following warnings appeared in my /var/log/messages: May 28 11:23:34 rpc-linux udevd-work[6414]: device node '/dev/mapper/raid5-usrsrc' already exists, link to '/dev/mapper/raid5-usrsrc' will not overwrite it May 28 11:13:37 rpc-linux udevd-work[6414]: device node '/dev/mapper/raid5-usrsrc' already exists, link to '/dev/mapper/raid5-usrsrc' will not overwrite it i am pretty sure device node did not exist! next, these warnings appeared twice (issue begins using kernel 2.6.37 and newer) during boot: the first lvm2 checks and the seconds local file system checks and appeared for every lvm partition - in my case: # lvs LV VG Attr LSize Origin Snap% Move Log Copy% Convert ccache raid5 -wi-ao 4.00g distfiles raid5 -wi-ao 6.00g home raid5 -wi-ao 100.00g media raid5 -wi-ao 1.98t opt raid5 -wi-ao 4.00g portage raid5 -wi-ao 1.00g swap raid5 -wi-ao 8.00g tmp raid5 -wi-ao 4.00g usr raid5 -wi-ao 10.00g usrsrc raid5 -wi-ao 8.00g var raid5 -wi-ao 4.00g vartmp raid5 -wi-ao 8.00g please, advice me a workaround or a fix. it is not problem to give us more info, it is enough to ask what do you need and i will cooperate with pleasure... thanks --JB
(In reply to comment #11) please remove this one and #11, thanks
