Red Hat Bugzilla – Bug 177411
LVM scans physical paths after mpio is set up
Last modified: 2007-11-30 17:07:22 EST
Description of problem: LVM2 tools scan and use physical paths even after mpio is set up. Version-Release number of selected component (if applicable): RHEL 4 U3 Beta How reproducible: Always Steps to Reproduce: 1. Set up device-mapper-multipath 2. Create lvm volume on multipath device 3. run any lvm command that scans (e.g. vgdisplay) Actual results: [root@clu1 ~]# vgdisplay Found duplicate PV tExTG5xhMixAeJF3xsAmTb2mtciOwU0Y: using /dev/sde not /dev/sdb Found duplicate PV u8Phs1yb0Cx2xJhiUSG1P27YmF5ILIJH: using /dev/sdg not /dev/sdd Expected results: No errors Additional info: There are several ways to solve this. One would be to change LVM scanning functions to scan dm devices first, and ignore subsequent duplicate PVs found on the physical paths. Another way would be for the device-mapper-multipath tools to add a filter entry in lvm.conf to ignore the devices that are configured when multipath is first run.
Please supply: lvm.conf file output of 'lvm version' output of 'vgscan -vvvv'
and also output of 'dmsetup info -c' and 'dmsetup table' (to show the multipath config)
(my suspicion is that it's just the error message that needs fixing, but the code is actually doing the right thing)
turns out that this might only be an issue when the LUN already has the LVM volume. For example, reinstalling the OS on the server. So what is probably happening is that the system scans (at boot) the physical paths. Later, the user logs in and configures multipathing. So in theory, a reboot might solve this. I am testing now (reinstalling).
Ok so the issue is relatively minor. Only happens when multipath is not set up at boot, and then gets set up. Reboot fixes it. Is there a way to fix this short of a reboot?
Without the diagnostics I asked for, I still don't know what the problem is.
i.e. I don't know if the tools were or were not correct in displaying those error messages.