From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Rajnoha Date: Tue, 01 Oct 2013 13:24:59 +0200 Subject: lvmetad activation problem with MD devices In-Reply-To: <524A744D.207@redhat.com> References: <20130930203615.59bfd346@work.puleglot> <524A744D.207@redhat.com> Message-ID: <524AB10B.2050705@redhat.com> List-Id: To: lvm-devel@redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit On 10/01/2013 09:05 AM, Peter Rajnoha wrote: > On 09/30/2013 06:36 PM, Alexander Tsoy wrote: >> Commit 8d1d83504dcf9c86ad42d34d3bd0b201d7bab8f6 introduced the >> following problem. If MD device is assembled in initramfs and some LVs >> on it are not activated, then those LVs still not activated during >> system boot. > > Yes, sorry for the problem. I'm just working on a fix. The source of > the problem here is that udev database is not handed over from initramfs > for MD devices and so the udev state is simply lost. The state information > we need is the MD activation state. There's also a bug reported for this issue: https://bugzilla.redhat.com/show_bug.cgi?id=1011704 I've handed that over to dracut as we can solve this by adding a special 'OPTIONS+="db_persist"' udev rule to MD rules that are installed for MD devices in initramfs. This is already done for device-mapper devices in dracut. The proposed patch is simple (also tested on my system where I reproduced the original problem): https://bugzilla.redhat.com/attachment.cgi?id=805841 Peter