raid
in [Slackware]
|
Prev: most popular/reliable remote desktop control software
Next: Canadian Presence requirements for .CA domains (Was: Justlinux.ca)
From: Douglas Mayne on 3 Nov 2009 21:20 I am trying to fill in some of the gaps in my knowledge. I want to learn the ins and outs of RAID setup with mdadm. I started with a simple test: a 20G raid 1 (mirror) on separate sata disks and partitions. I partitioned the physical devices in advance with fdisk: root(a)atom-330:~# fdisk /dev/sda -l Disk /dev/sda: 1000.2 GB, 1000204886016 bytes 255 heads, 63 sectors/track, 121601 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x57c89da2 Device Boot Start End Blocks Id System /dev/sda1 1 262 2104483+ 83 Linux /dev/sda2 263 2874 20980890 fd Linux raid autodetect root(a)atom-330:~# fdisk /dev/sdb -l Disk /dev/sdb: 1000.2 GB, 1000204886016 bytes 255 heads, 63 sectors/track, 121601 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x66709e1c Device Boot Start End Blocks Id System /dev/sdb1 1 262 2104483+ 83 Linux /dev/sdb2 263 2874 20980890 fd Linux raid autodetect Then I created the array: # mdadm --create /dev/md0 --level=mirror --raid-devices=2 /dev/sda2 /dev/sdb2 root(a)atom-330:~# mdadm --detail /dev/md0 /dev/md0: Version : 00.90.03 Creation Time : Wed Nov 4 07:56:25 2009 Raid Level : raid1 Array Size : 20980800 (20.01 GiB 21.48 GB) Used Dev Size : 20980800 (20.01 GiB 21.48 GB) Raid Devices : 2 Total Devices : 2 Preferred Minor : 0 Persistence : Superblock is persistent Update Time : Wed Nov 4 09:47:33 2009 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 UUID : 2bbba569:5e093638:8e7a9340:a94dfb09 Events : 0.1698 Number Major Minor RaidDevice State 0 8 2 0 active sync /dev/sda2 1 8 18 1 active sync /dev/sdb2 The array is used as the root filesystem, and is initialized with the sequence in the initial ramdisk. # mdadm -E -s >/etc/mdadm.conf # mdadm -A -s When the device /dev/md0 is started and mounted it generates these messages to the ring buffer, dmesg: <snippet of dmesg output> md: md0 stopped. md: bind<sdb2> md: bind<sda2> raid1: raid set md0 active with 2 out of 2 mirrors md0: unknown partition table XFS mounting filesystem md0 Ending clean XFS mount for filesystem: md0 <end dmesg output> Has anyone seen this error/warning: "md0: unknown partition table." I googled a bit for it, and I see it comes up quite often, but I can't tell for sure if it is something that I should be worried about, or not. TIA p.s. I am running a self compiled 2.6.30.9 kernel.
From: Grant Taylor on 4 Nov 2009 00:11 On 11/3/2009 8:20 PM, Douglas Mayne wrote: > I am trying to fill in some of the gaps in my knowledge. I want to > learn the ins and outs of RAID setup with mdadm. :) > The array is used as the root filesystem, and is initialized with the > sequence in the initial ramdisk. Can I ask why you are using an initial RAM disk? If you are using the RAID Auto Detect partition type, the kernel should see and activate the RAID (if it can) auto-magically on boot. With the RAID activated, you can simply mount your root directly off of it. > Has anyone seen this error/warning: "md0: unknown partition table." I > googled a bit for it, and I see it comes up quite often, but I can't > tell for sure if it is something that I should be worried about, or > not. I believe that is simply the kernel examining partition structures on block devices that it sees. Seeing as how you are likely putting a file system directly on to the md device, there is no partition structure there to be found. I do not think this is any thing that you need to be worried about. I have heard that it is possible to get partition support on md devices (kernel option?) but I have never done it. Grant. . . .
From: Douglas Mayne on 4 Nov 2009 06:58 On Tue, 03 Nov 2009 23:11:40 -0600, Grant Taylor wrote: > On 11/3/2009 8:20 PM, Douglas Mayne wrote: >> I am trying to fill in some of the gaps in my knowledge. I want to >> learn the ins and outs of RAID setup with mdadm. > > :) > >> The array is used as the root filesystem, and is initialized with the >> sequence in the initial ramdisk. > > Can I ask why you are using an initial RAM disk? > I am using a modular kernel, similar to the standard Slackware modular kernel. I need the initial ramdisk to load the XFS module. I think it will come in handy for my next exercise: root filesystem on iSCSI. > > If you are using the RAID Auto Detect partition type, the kernel should > see and activate the RAID (if it can) auto-magically on boot. With the > RAID activated, you can simply mount your root directly off of it. > >> Has anyone seen this error/warning: "md0: unknown partition table." I >> googled a bit for it, and I see it comes up quite often, but I can't >> tell for sure if it is something that I should be worried about, or >> not. > > I believe that is simply the kernel examining partition structures on > block devices that it sees. Seeing as how you are likely putting a file > system directly on to the md device, there is no partition structure > there to be found. I do not think this is any thing that you need to be > worried about. > ok. > > I have heard that it is possible to get partition support on md devices > (kernel option?) but I have never done it. > I looked at the man page a little closer. It includes this info: --auto{=no,yes,md,mdp,part,p}{NN} Instruct mdadm to create the device file if needed, possibly allocating an unused minor number. "md" causes a non-partition- able array to be used. "mdp", "part" or "p" causes a partition- able array (2.6 and later) to be used. I am guessing that partition support would allow mirroring of entire disks (i.e. /dev/sda and /dev/sdb in my case). This could be useful, and more like hardware raid. > > > Grant. . . . > So far, I am impressed with the features of mdadm. I also tested building with one device "missing," and then adding the second device. I tracked the status as obtained from /proc/mdstat. That was quick- under 10min for a 20G partition. -- Douglas Mayne
From: Grant Taylor on 4 Nov 2009 20:46
On 11/4/2009 5:58 AM, Douglas Mayne wrote: > --auto{=no,yes,md,mdp,part,p}{NN} Instruct mdadm to create the device > file if needed, possibly allocating an unused minor number. "md" > causes a non-partition-able array to be used. "mdp", "part" or "p" > causes a partition-able array (2.6 and later) to be used. Good to know. I'll have to give the partitioning support a try some time. > I am guessing that partition support would allow mirroring of entire > disks (i.e. /dev/sda and /dev/sdb in my case). This could be useful, > and more like hardware raid. You can mirror an entire block device with out the partitioning support. Though I'd advise against it because the partitions aid when having to do data recovery on the drive (most utilities only know how to work with in a partition like structure). You would also see a similar error for the lack of partitions on your raw /dev/sda & /dev/sdb devices like you are seeing on /dev/md0 device. > So far, I am impressed with the features of mdadm. I also tested > building with one device "missing," and then adding the second > device. I tracked the status as obtained from /proc/mdstat. That was > quick- under 10min for a 20G partition. mdadm is *VERY* capable. There are a number of inexpensive raid card out there that run (an older?) Linux kernel and mdadm to do the RAID. You can also do some very interesting things. Create a RAID 1 with a missing disk, add the spare disk some day, and later convert the RAID 1 to a 3 disk RAID 5. You can even grow the RAID 5 down the road. (Converting from a non-RAID to a RAID 1 is not nice though.) Grant. . . . |