Red Hat Bugzilla – Bug 124647
Unable to use LVM on EMC PowerPath devices
Last modified: 2007-11-30 17:07:02 EST
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040429 Firefox/0.8 Description of problem: We have been attempting to make use of the Linux Logical Volume Manager on EMC PowerPath pseudo-devices. We are using the 3.0.6 version of the PowerPath driver on an EMC Clariion storage appliance. An attempt to create a physical volume on a PowerPath device fails with an invalid device error. Version-Release number of selected component (if applicable): kernel-smp-2.4.21-9.0 How reproducible: Always Steps to Reproduce: 1.Create EMC PowerPath devices 2.Attempt to create an LVM physical volume using pvcreate Actual Results: Invalid device error Expected Results: Success of the physical volume creation Additional info: We have been advised by EMC that what we are doing is unsupported by the driver, but we are wondering what other options are available. Do other Red Hat ES/AS versions suffer from this difficulty? Are other volume management solutions usable (e.g. IBM EVMS or others)?
It's probably better if you file a support request for this and not a bugzilla bug. our support people have the infrastructure to escalate issues with the EMC binary only kernel modules to EMC, we in engineering can't really do that with bugzilla.
Covered in lvm-1_0_8-2
An errata has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2004-232.html
Encountered same issue after updating to lvm-1.0.8-5 pvcreate -- invalid physical volume "/dev/emcpowera"
Me also: # rpm -q lvm lvm-1.0.8-5 # ls /dev/emcpowerb* /dev/emcpowerb /dev/emcpowerb12 /dev/emcpowerb2 /dev/emcpowerb6 /dev/emcpowerb1 /dev/emcpowerb13 /dev/emcpowerb3 /dev/emcpowerb7 /dev/emcpowerb10 /dev/emcpowerb14 /dev/emcpowerb4 /dev/emcpowerb8 /dev/emcpowerb11 /dev/emcpowerb15 /dev/emcpowerb5 /dev/emcpowerb9 # fdisk -l /dev/emcpowerb Disk /dev/emcpowerb: 68.3 GB, 68325212160 bytes 255 heads, 63 sectors/track, 8306 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/emcpowerb1 1 8306 66717913+ 8e Linux LVM # pvcreate /dev/emcpowerb pvcreate -- invalid physical volume "/dev/emcpowerb" Also note that lvmdiskscan does not see the emcpower psuedo devices either (probably the same issue): # lvmdiskscan lvmdiskscan -- reading all disks / partitions (this may take a while...) lvmdiskscan -- /dev/sda1 [ 101.94 MB] free whole disk lvmdiskscan -- /dev/sdb1 [ 399.62 GB] free whole disk lvmdiskscan -- /dev/sdc1 [ 63.63 GB] free whole disk lvmdiskscan -- /dev/sdd1 [ 336 GB] free whole disk lvmdiskscan -- /dev/sde1 [ 801.88 GB] free whole disk lvmdiskscan -- /dev/sdf1 [ 764.60 GB] free whole disk lvmdiskscan -- /dev/sdg1 [ 399.62 GB] free whole disk lvmdiskscan -- /dev/sdh1 [ 63.63 GB] free whole disk lvmdiskscan -- /dev/sdi1 [ 336 GB] free whole disk lvmdiskscan -- /dev/sdj1 [ 801.88 GB] free whole disk lvmdiskscan -- /dev/sdk1 [ 764.60 GB] free whole disk lvmdiskscan -- 11 disks lvmdiskscan -- 11 whole disks lvmdiskscan -- 0 loop devices lvmdiskscan -- 0 multiple devices lvmdiskscan -- 0 network block devices lvmdiskscan -- 15 partitions lvmdiskscan -- 0 LVM physical volume partitions
And same over here..doesn't work.
I think I have a possible fix: It seems like the patch makes power2 an invalid parition type as tested by the switch/case section tools/lib/lvm_check_partitioned.c else { switch (lvm_get_device_type(st_rdev)) { case LVM_DEVICE_TYPE_INVALID: case LVM_DEVICE_TYPE_NBD: case LVM_DEVICE_TYPE_LOOP: case LVM_DEVICE_TYPE_MD: case LVM_DEVICE_TYPE_ATARAID: case LVM_DEVICE_TYPE_POWER2: break; The case LVM_DEVICE_TYPE_POWER2: line simply needs to be removed and all is rosy. In diff form, something like this: --- lvm_check_partitioned_dev.c 2004-09-26 17:45:51.000000000 +0100 +++ lvm_check_partitioned_dev_ok.c 2004-09-26 17:47:58.000000000 +0100 @@ -65,7 +65,6 @@ case LVM_DEVICE_TYPE_LOOP: case LVM_DEVICE_TYPE_MD: case LVM_DEVICE_TYPE_ATARAID: - case LVM_DEVICE_TYPE_POWER2: break; default: ret = TRUE;
We just installed PowerPath 3.0.7. The problem still exists!
I have the same problem while using PowerPath 3.0.7 with lvm-1.8.0-5 on Red Hat AS 3.0. It doesn't work for me. I have stuck at this point :-(. pvcreate -- invalid physical volume "/dev/emcpowera" Will be this issue fixed soon?. Thanks.
The patch described in comment #8 fixes the problem. I have verified this with PowerPath 3.0.6.
[root@eabodb01 root]# rpm -q lvm lvm-1.0.8-5 [root@eabodb01 root]# ls /dev/emcpowera* /dev/emcpowera /dev/emcpowera11 /dev/emcpowera14 /dev/emcpowera3 /dev/emcpowera6 /dev/emcpowera9 /dev/emcpowera1 /dev/emcpowera12 /dev/emcpowera15 /dev/emcpowera4 /dev/emcpowera7 /dev/emcpowera10 /dev/emcpowera13 /dev/emcpowera2 /dev/emcpowera5 /dev/emcpowera8 [root@eabodb01 root]# fdisk -l /dev/emcpowera Disk /dev/emcpowera: 107.3 GB, 107374182400 bytes 255 heads, 63 sectors/track, 13054 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/emcpowera1 1 13054 104856223+ 8e Linux LVM [root@eabodb01 root]# pvcreate /dev/emcpowera pvcreate -- invalid physical volume "/dev/emcpowera" [root@eabodb01 root]# pvcreate /dev/emcpowera1 pvcreate -- invalid physical volume "/dev/emcpowera1" [root@eabodb01 root]# lvmdiskscan lvmdiskscan -- reading all disks / partitions (this may take a while...) lvmdiskscan -- /dev/sda1 [ 100 GB] free whole disk lvmdiskscan -- /dev/sdb1 [ 100 GB] free whole disk lvmdiskscan -- /dev/sdc1 [ 100 GB] free whole disk lvmdiskscan -- /dev/sdd1 [ 100 GB] free whole disk lvmdiskscan -- /dev/cciss/c0d0p1 [ 94.10 MB] free whole disk lvmdiskscan -- 5 disks lvmdiskscan -- 5 whole disks lvmdiskscan -- 0 loop devices lvmdiskscan -- 0 multiple devices lvmdiskscan -- 0 network block devices lvmdiskscan -- 6 partitions lvmdiskscan -- 0 LVM physical volume partitions I agree, it does not seem to work. EMC has been of little help with this problem.
Sorry Matt, it took EMC so long... If the package won't arrive in the next 1-2 days, nothing will happen before January because of my XMas vacation anyway.
shouldnt this bug be reopened since the fix didnt work?
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHEA-2004-589.html
I'm having the exact problem on a newly installed system running lvm-1.0.8-9 on RHES3?! Is there a patch, etc.?
Fix will be part of U5.
(In reply to comment #38) > Fix will be part of U5 There won't be an update prior??? I'm dead in the water now! Please, is there a patch, etc. I can apply now? How long until U5 is out?
I've verified the bug still exists with lvm-1.0.8-12.2 obtained from the beta channel.
For me, the problem was caused by LVM not recognizing the volume being called by the path /dev/emcpowera and/or /dev/emcpowera1. Being that I'm not clustering and I don't need the functionality provided by EMC's app/wrapper PowerPath, I uninstalled it and used the standard device name /dev/sdX.
(In reply to comment #39) > (In reply to comment #38) > > Fix will be part of U5 > There won't be an update prior??? I'm dead in the water now! Please, is there a > patch, etc. I can apply now? How long until U5 is out? I have made a fix that works for me. Looking in the code, i found the folow fix in lvm-1.0.8-9-power2.patch: --- LVM/1.0.8/tools/lib/lvm_check_partitioned_dev.c.power2 Wed Oct 20 10:51:20 2004 +++ LVM/1.0.8/tools/lib/lvm_check_partitioned_dev.c Wed Oct 20 10:53:33 2004 @@ -65,7 +65,6 @@ case LVM_DEVICE_TYPE_LOOP: case LVM_DEVICE_TYPE_MD: case LVM_DEVICE_TYPE_ATARAID: - case LVM_DEVICE_TYPE_POWER2: break; default: ret = TRUE; @@ -165,6 +164,7 @@ { "ataraid", LVM_DEVICE_TYPE_ATARAID }, /* ATA Raid */ { "viod", LVM_DEVICE_TYPE_VIODASD }, /* viodasd */ { "i2o_block", LVM_DEVICE_TYPE_I2O },/* i2o disk */ + { "emcpower", LVM_DEVICE_TYPE_POWER2 },/* PowerPath */ { "power2", LVM_DEVICE_TYPE_POWER2 },/* PowerPath */ { "emd", LVM_DEVICE_TYPE_EMD }, /* Enhanced MD */ { "sx8", LVM_DEVICE_TYPE_SX8 }, /* Promise SX8 SATA */ The complete code read the /proc/devices (like: cat /proc/devices ) and check every device in this list if they match with the above array. The situation is that in /proc/devices the powerpath show as "emcp" instead of emcpower, My fix is change this: + { "emcpower", LVM_DEVICE_TYPE_POWER2 },/* PowerPath */ to this: + { "emcp", LVM_DEVICE_TYPE_POWER2 },/* PowerPath */ My rebuild work for partitions (like: emcpowerb1) but don't work for disks emcpowerb But i don't see any problem in creating a entire disk partition type 8e (Linux LVM)
Any news on this ? It is not fixed in lvm-1.0.8-12.2 (Update 5). We've hand patched using the fix from the comment #66, but we'd like an official/supported/easily installable fix.
Pierre, the fix from #66 will be included in the next release.
Yes, RHEL 3, Update 2 will have this fix.
I meant, RHEL 3, Update 6
*** Bug 113971 has been marked as a duplicate of this bug. ***
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2005-579.html