All of lore.kernel.org
 help / color / mirror / Atom feed
* mdraid + dm-multipath - non-dm-mulitpath boot md coexisting with dm-multi md data volume?
@ 2015-10-01 18:55 Jeff Johnson
  0 siblings, 0 replies; only message in thread
From: Jeff Johnson @ 2015-10-01 18:55 UTC (permalink / raw)
  To: linux-raid

Greetings,

I'm looking for advice on configuring mdadm.conf to allow for the 
coexisting of a non dm-multipath boot/root md device and a dm-multipath 
configured md device data volume.

The issue I am running into is that on boot mdraid assumes control of 
all disks with md superblocks on them before multipath starts. This 
appears to be due to the boot/root being md devices, multipath never 
gets to own the other disks in the system and then assemble on mpath 
devices in /dev/disk/by-id.

I have tried using DEVICE declarations in mdadm.conf but so far the 
result is either the boot volumes don't get recognized (assemble + 
switchroot fails) but the data volume gets assembled with dm-multipath 
devices, or the boot/root md devices assemble and the dm-multipath based 
md devices get assembled as sd member devices and dm-multipath is disabled.

Some of the DEVICE declarations I tried in mdadm.conf are below. Two on 
the same line or two separate lines. The non dm-mulitpath boot devices 
are Intel SATA SSDs on the onboard SATA ports. The dm-mulitpath RAID-10 
data volume is on an external SAS HBA with valid multipath 
configuration. I have remade the initramfs after these changes (dracut -f).

DEVICE /dev/disk/by-id/scsi-SATA_INTEL_SSD.*-.* 
/dev/disk/by-id/dm-uuid-.*-mpath-.*

and

DEVICE /dev/disk/by-id/scsi-SATA_INTEL_SSD.*-.*
DEVICE /dev/disk/by-id/dm-uuid-.*-mpath-.*

These examples find the SAS attached dm-multipath devices fine and 
assemble the mdraid volumes but the boot devices are not found or 
assembled. A look via rdshell shows the 
/dev/disk/by-id/scsi-SATA_INTEL_SSD.*-.* devices listed in the 
/dev/disk/by-id directory. It appears mdadm doesn't like the declaration 
for the SSDs and won't find them. I haven't been able to figure out why 
so far.

Current state (/proc/mdstat)
cat /proc/mdstat
Personalities : [raid1] [raid10]
md4 : active raid10 dm-11[3] dm-5[2] dm-7[1] dm-9[0]
       584615936 blocks super 1.2 512K chunks 2 near-copies [4/4] [UUUU]
       bitmap: 0/5 pages [0KB], 65536KB chunk

md3 : active raid1 dm-10[3] dm-2[2] dm-4[1] dm-8[0]
       521792 blocks super 1.2 [4/4] [UUUU]

md0 : active raid1 sdb1[1] sda1[0]
       255936 blocks super 1.0 [2/2] [UU]

md2 : active raid1 sdb3[1] sda3[0]
       225653760 blocks super 1.1 [2/2] [UU]
       bitmap: 0/2 pages [0KB], 65536KB chunk

md1 : active raid1 sda2[0] sdb2[1]
       8380416 blocks super 1.1 [2/2] [UU]

unused devices: <none>

OS: CentOS 6.6 x86_64 2.6.32-504.el6.x86_64

If anyone has suggestions of how to configure a proper set of DEVICE 
declarations to assemble the SATA connected member drives of the 
boot/root volumes without relying on the default "DEVICE partitions 
containers" where the multipath devices get assembled as sd devices 
prior to multipath starting I would appreciate the advice.

Thanks,

--Jeff

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2015-10-01 18:55 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-10-01 18:55 mdraid + dm-multipath - non-dm-mulitpath boot md coexisting with dm-multi md data volume? Jeff Johnson

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.