LinuxQuestions.org - Fedora 5 raid setup: recommendations and help

- Linux - Newbie (https://www.linuxquestions.org/questions/linux-newbie-8/)

- - Fedora 5 raid setup: recommendations and help (https://www.linuxquestions.org/questions/linux-newbie-8/fedora-5-raid-setup-recommendations-and-help-460078/)

Fedora 5 raid setup: recommendations and help

I'm setting up Fedora 5 in an almost identical setup to my last box ... except for raid. I have an old Athalon 800 with 512Mb of PC133 in it that I'm using as an in-house web server. Of course, that didn't come with SATA capability or raid capability, so now that we have some data sil_agageidcdccethat needs some kind of preserving, I'm looking to add SATA Raid 1. I got a card that has linux support (no problem, really, a 2 SATA port PCI card and two 120Gb drives.

Now, I'm planning the following:
root and boot will be on my IDE drive (a 40Gb Western Digital). I'll store a /data mount on the mirrored SATA drives.

The questions come in now:
First, hardware vs software? In the install, Linux picks up the hardware raid and shows it as /dev/mapper/sil_agageidcdcce ... strange name, but what do I care. It shows as a single 120Gb device. In this case, I put 1Gb of swap also on the IDE drive. I'm willing to try again, but last time I tried to set up this configuration, the installer froze on disk 2 (of 5 ... come on, do you have to add a disk for each new release?).

I also tried software, and I admit, I tried something trickier than I probably should have. Reading the how-to's and manuals, it seemed like it was possible in software raid to actually have two raid devices on one pair of drives. In this case, I thought, hey, I'll set up Raid 0 for 1Gb of swap, and then 119Gb of Raid 1 for the /data directory. I did as directed in the manuals, I created a raid device with 1Gb of swap, 119Gb of /data, and then mirrored it onto the other drive. All was going well until the installer (Anaconda, I believe) started formatting the raid device. Then it froze.

Right now, I gave up on the raid, thinking there might be a way to add it later, so I installed Fedora on the IDE drive with nothing at all on the SATA drives right now. Both are visible (/dev/sda and sdb), and even the hardware /dev/mapper/sil-thingy is still visible (I turned the hardware raid back on). However, being somewhat new to this (I'll blame it on that), I can't seem to get the drive to mount. I could probably mount sda and sdb independently, but then they wouldn't be raided(?).

At this point, I'm at somewhat of a loss. I can't find an example out there that seems to work. Any help would be greatly appreciated.

personally i'd suggest software raid. there si very little performance hit compared to that you'd expect from a hardware one, and you still get a lot more control directly in the operating system. just create a partition on each device (note you'd mount /dev/sda1 not /dev/sda etc...) and use mdadm to wop it into a /dev/md0. a good tutorial would be right here... http://www.linuxdevcenter.com/pub/a/...2/05/RAID.html

I'll give it a shot ... might be a minute before I respond, though. I keep getting the error that sda2 (I'm going to try both raid 1 and raid 0 with mdadm) is already in use by the system ... which is plain rediculous because it's just a large unformatted glob so far. Ah well, I'm on a fresh install anyway, so I'll just reinstall and partition it "correctly" from the beginning. I'll tell you how it goes!

So I'm continuing to have issues, and wondering now where to resolve them. I went through the re-install and was using the installer to set up partitions. I set them up as follows:

hda: 100Mb /boot
hdb: 18Gb /
sda1: 1Gb
sdb1: 1Gb
Raid 0: sda1, sda2 /swap
sda2: 119Gb
sdb2: 119Gb
Raid 1: sda2, sdb2 /data

It hangs on formatting /data ... just locks up, no errors. Am I missing something?

Actually, I'm beginning to suspect the SATA drives ... my only successfull install was done when I ignored them completely

I've gotten further (I think). My fdisk -l now looks like this:

Code:

Disk /dev/hda: 20.0 GB, 20020396032 bytes

255 heads, 63 sectors/track, 2434 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes



  Device Boot      Start        End      Blocks  Id  System

/dev/hda1  *          1          13      104391  83  Linux

/dev/hda2              14        2434    19446682+  8e  Linux LVM



Disk /dev/sda: 120.0 GB, 120034123776 bytes

255 heads, 63 sectors/track, 14593 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes



  Device Boot      Start        End      Blocks  Id  System

/dev/sda1  *          1        131    1052226  fd  Linux raid autodetect

/dev/sda2            132      14593  116166015  fd  Linux raid autodetect



Disk /dev/sdb: 120.0 GB, 120034123776 bytes

255 heads, 63 sectors/track, 14593 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes



  Device Boot      Start        End      Blocks  Id  System

/dev/sdb1  *          1        131    1052226  fd  Linux raid autodetect

/dev/sdb2            132      14593  116166015  fd  Linux raid autodetect



Disk /dev/md0: 2154 MB, 2154823680 bytes

2 heads, 4 sectors/track, 526080 cylinders

Units = cylinders of 8 * 512 = 4096 bytes



Disk /dev/md0 doesn't contain a valid partition table



Disk /dev/md1: 237.9 GB, 237907738624 bytes

2 heads, 4 sectors/track, 58082944 cylinders

Units = cylinders of 8 * 512 = 4096 bytes



Disk /dev/md1 doesn't contain a valid partition table

I did everything in bootup except /dev/md1, /dev/sda2, and /dev/sdb2 ...

Here's the interesting part (to me). I have /no/ idea how fedora got it into it's head to make md1 raid0. I made md0 a raid0, but was hoping to make md1 a raid1.

Another interesting part, I try to mount /dev/md1 to something (like /mnt/data) and I get an error ... first telling me to specify the filesystem type, so I mount -t ext3 /dev/md1 /mnt/data, and I get an error:

Code:

 wrong fs type, bad option, bad superblock on /dev/md1,

      missing codepage or other error

      In some cases useful info is found in syslog - try

      dmesg | tail  or so

I check the dmesg file and it just says it can't find ext3. Fair enough, I never formatted the sucker (and honestly don't know how ... do I format sda2 and sdb2 separately or md1 ... and with what util?)

So I'm stuck again, but stuck in a better place.

Any help would be appreciated.

you would format md1 - "mke2fs -j /dev/md1" will give you an ext3 filesystem.

Well, I thought I'd try formatting just to see, but it gets hung.

It goes to block 137/1773 in creating the inode tables, and then just stops.

Is this indicative of a problem with the disks?

Secondarily, I still don't know where or how to change md1 from a raid 0 device to a raid 1 device.

I keep seeing the following error in messages:

Code:

Jul  2 12:26:44 localhost kernel: ata2: command 0x35 timeout, stat 0xd9 host_stat 0x61

Jul  2 12:26:44 localhost kernel: ata2: status=0xd9 { Busy }

Jul  2 12:26:44 localhost kernel: sd 1:0:0:0: SCSI error: return code = 0x8000002

Jul  2 12:26:44 localhost kernel: sdb: Current: sense key: Aborted Command

Jul  2 12:26:44 localhost kernel:    Additional sense: Scsi parity error

Jul  2 12:26:44 localhost kernel: end_request: I/O error, dev sdb, sector 2749187

Jul  2 12:26:44 localhost kernel: ATA: abnormal status 0xD9 on port 0xE0832CC7

And mke2fs won't die.

The more I look around, the more it looks like a problem with the driver, actually ... I'm using a Syba SD-Sata3112-150R (which claims linux support, and since the drives are recognized, I've got to say it's probably right). I found this on linuxforums:

http://www.linuxforums.org/forum/sla...-too-well.html

That suggests it might be a problem with this scsi_set_device in the driver ... now if I just knew how to edit drivers (specifically ... where the driver for the card is located).

OK, well if a patch is ever found to fix a kernel driver, they are normally updated fairly quickly, but then fc5 hasn't been about *that* long really... still, run a full yum update if possible and get the very latest kernel installed, and that may well be patched against it. google says little about it and i don't know engouh about the problem to dig into in in the kernel changes either. if that doesn't fix it then you'd want to look at patching the kernel source yourself, but if this is the *only* change you want to make to an otherwise stock kernel it's actually very simple and safe.

as far as changing your raid device, that is an option you need to specify as and when you build the inital array, you can't convert from one to another.

Quote:

as far as changing your raid device, that is an option you need to specify as and when you build the inital array, you can't convert from one to another.

That's part of what I'm trying to figure out ... I never specified a raid for sda2 and sdb2 ... I just used fdisk to make them linux software raid partitions (I think it says linux raid autodetect), and then I rebooted so I could try using mdadm on it ... suddenly I had a giant unformatted raid 0 array that didn't seem to be declared anywhere and couldn't be mounted or formatted.

I've also now tried reinstalling, this time with the drives simply as individual drives. When it tries to partition them and format them ... it hangs. This is looking more like an issue of linux recognizing these drives or the drives being bad (back to the controller card again). Maybe I should take this to the hardware forum.