From: Petr Rockai Date: Tue, 30 Nov 2010 12:12:47 +0100 Subject: [PATCH] Disable automatic MISSING_PV recovery when handles_missing_pvs is set (BZ 653643) In-Reply-To: <20101129224015.GS29380@agk-dp.fab.redhat.com> (Alasdair G. Kergon's message of "Mon, 29 Nov 2010 22:40:15 +0000") References: <87d3ptm965.fsf@twilight.int.mornfall.net.> <20101129224015.GS29380@agk-dp.fab.redhat.com> Message-ID: <87hbezjc9c.fsf@twilight.int.mornfall.net.> List-Id: To: lvm-devel@redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Hi, Alasdair G Kergon writes: > On Thu, Nov 25, 2010 at 09:37:38PM +0100, Peter Rockai wrote: >> it seems I did everything but actually posted the proposed patch for BZ >> 653643. The explanation of what goes wrong where and how the patch fixes >> that is in BZ comments. > > So what does the new output in the test case in that bz look like? > > Do we still get the last WARNING message? (Where does it come from? > Maybe it's too hard to fix that without removing it from other code > paths where we do still want it?) Which WARNING message you mean? There is only one in each command: inconsistent metadata, we can't get rid of that one. About the PV not found, I think it's better with the message than without. On the other hand, we *could* make vgextend --restoremissing behave just like normal vgextend if the PV is actually not part of the VG. But I think it's better for scriptability (and predictability) to not do that. Yours, Petr