From: "Markus Baertschi" In-Reply-To: <20030910132010.GC10632@percy.comedia.it> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: Re: [linux-lvm] Devices seen twice - how does LVM handle ? Message-Id: Sender: linux-lvm-admin@sistina.com Errors-To: linux-lvm-admin@sistina.com Reply-To: linux-lvm@sistina.com List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: Date: Wed Sep 10 11:29:02 2003 List-Id: To: "linux-lvm@sistina.com" Hi=20Luca, On=20Wed,=2010=20Sep=202003=2015:20:11=20+0200,=20Luca=20Berra=20wrote: >>However,=20does=20LVM=20handles=20this=20situation=20(disks=20visible=20= twice)=20nicely=20? >does=20not=20seem=20broken=20to=20me=20if=20you=20have=20two=20hba. >lvm=20does=20not=20support=20multipath What=20does=20LVM=20do=20if=20it=20sees=20two=20paths=20(/dev/sdx1=20entri= es)=20to=20the=20same virtual=20disk=20? >>or=20should=20I=20continue=20to=20use=20my=20workaround=20(pull=20a=20fi= ber)=20to=20make=20sure >>the=20disks=20are=20visible=20only=20once=20? >it=20is=20not=20a=20nice=20workaround. Yes,=20but=20the=20only=20one=20I=20got=20to=20work=20in=20the=20timeframe= =20I=20had=20to=20get=20the machines=20up=20and=20running=20again. >>-=20Are=20there=20risks=20of=20corruption=20? >i=20don't=20think=20so. Why=20? >>-=20What=20happens=20if=20the=20path=20the=20LVM=20happens=20to=20use=20= breaks,=20but=20the=20other >>is=20still=20up.=20Will=20LVM=20handle=20fail-over=20? >lvm=20does=20not=20support=20multipath OK. >at=20the=20moment=20your=20best=20chance=20is=20setting=20up=20an=20md=20= multipath >and=20building=20your=20lvm=20over=20the=20md=20devices.=20be=20sure=20to= =20use=20mdadm=20instead >of=20raidtools=20to=20manage=20the=20md. I'm=20not=20familiar=20with=20md.=20The=20volumes=20are=20already=20there=20= and=20in=20production, If=20possible=20I=20don't=20want=20to=20reconfigure=20(downtime...). >i=20have=20this=20setup=20(2=20hbas=20to=20same=20storage=20with=20lvm=20= over=20md=20multipath)=20in >production,=20and=20it=20works=20fine. >another=20option=20could=20be=20if=20your=20hba=20driver=20handles=20mult= ipath.=20i=20know >some=20qlogic=20boards=20do. I'm=20using=20qlogic=20boards=20with=20the=20qla2300=20failover=20driver.=20= This=20was=20working perfectly=20until=20someone=20installed=20a=20blade=20center=20on=20unrela= ted=20ports of=20the=20same=20storage=20box=20and=20changed=20the=20WWN=20of=20the=20b= ox=20in=20the=20process. The=20linux=20servers=20were=20rebooted=20three=20months=20(!)=20later=20f= or=20an=20unrelated purpose=20and=20promptly=20failed=20to=20come=20up=20(vgscan=20segfault...= ).=20The=20failover driver=20now=20thinks=20we=20have=20two=20storage=20boxes... In=20the=20short=20run=20the=20only=20solution=20I=20had=20was=20to=20remo= ve=20one=20path/fiber. The=20real=20fix=20is=20to=20reconfigure=20the=20storage=20box,=20but=20th= is=20has=20to=20be=20done when=20I=20can=20stop=20all=20servers=20connected=20to=20it=20(sunday=2002= :00=20AM...). In=20the=20meantime=20I=20installed=20the=20latest=20LVM=20and=20this=20fi= xes=20the=20segfault problem.=20My=20main=20priority=20is=20stability=20right=20now=20and=20I=20= prefer=20running on=20one=20fiber=20than=20in=20a=20potentially=20dangerous=20configuration= .=20The=20fiber can=20be=20switched=20in=20a=20hurry=20(just=20plug=20in=20the=20other=20o= ne),=20software=20needs skills=20the=20local=20people=20don't=20have=20(I'm=20most=20of=20the=20ti= me=20not=20on=20site). Markus -- =20=20Markus=20Baertschi=20=20=20=20=20=20=20=20=20=20=20=20=20Phone:=20++= 41=20(21)=20807=201677 =20=20Bas=20du=20Ross=E9=2014b=20=20=20=20=20=20=20=20=20=20=20=20=20Fax=20= =20:=20++41=20(21)=20807=201678 =20=20CH-1163,=20Etoy=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20Email= :=20markus@markus.org =20=20Switzerland=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20Hom= epage:=20www.markus.org