Rick, perhaps your lvm.conf file needs a tweak to see the multipath lun device?
http://support.novell.com/techcenter/sdb/en/2005/04/sles_multipathing.ht ml "How to setup / use multipathing on SLES" in section 4 discusses "Using LVM2 on top of the MPIO devices" and the need to adjust 'filter =' and 'types =' statements. Depending on your site's setup you might have to go a step beyond what Novell says and add /dev/mapper to the filter too. filter = [ "a|/dev/dasd.*|","a|/dev/mapper.*|","a|/dev/disk/by-name/.*|","r|.*|"] -------------------------------------------------------- This e-mail, including any attachments, may be confidential, privileged or otherwise legally protected. It is intended only for the addressee. If you received this e-mail in error or from someone who was not authorized to send it to you, do not disseminate, copy or otherwise use this e-mail or its attachments. Please notify the sender immediately by reply e-mail and delete the e-mail from your system. -----Original Message----- From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of Richard Troth Sent: Wednesday, April 25, 2007 1:14 AM To: LINUX-390@VM.MARIST.EDU Subject: LVM -vs- device mapper We were making progress with SAN ... and now we're stuck again. Actually, SAN itself is working pretty reliably at this point. But I cannot figure out how to feed a multipath SAN volume into LVM2. In VM things, you cannot connect to the virtual console of a virtual machine hosting an LDEV. The point is, you should not sign on to TCPIP on a logical terminal hosted by TCPIP. The test is simplistic and we have learned to live with it. (You can get around it by dialing thru VTAM, for example.) The test fails goes beyond real protection in that you can also not sign on to TCPIP1 from an LDEV hosted by TCPIP2, which should be allowed. But this is just for example. The real problem has nothing to do with terminal logical devices. It's a SAN and logical volume problem. I BELIEVE that LVM has a certain belt-and-suspenders check with a similar nature. Is there a way around it? Specifically, if I create a multipath volume I cannot then feed that to LVM as a PV. If I masquerade the multipath logical volumes each behind a loopback (pseudo block device) then it works: I can make it be a PV and can build a volume group on that with its own logical volumes. How can I do this without the hack?? The multipath LUN appears to LVM as if it were an LV. (In some reality, it is an LV, but not an LV of LVM2's creation, so this should be allowed.) Maybe someone can enlighten me as to how to do this with LVM alone. Would rather leave the two parts separate. I don't see from the LVM doc how to feed two paths as one PV. I had some luck with EVMS, but it changes the game in ways that we cannot do at this time. If I could just get 'pvcreate' to not reject a logical-for-another-reason LV as a PV then this would work. Sample follows: /dev/sda and /dev/sdb are 'dmsetup' combined into /dev/mapper/mylun. pvcreate /dev/mapper/mylun fails with "Device /dev/mapper/mylun not found." losetup /dev/loop5 /dev/mapper/mylun pvcreate /dev/loop5 works! -- R; ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
