12.04, 12.10 x64 Server install problems - swap or RAID-related?

[jeffmefun]

I'm having problems installing Ubuntu Server similar to those I've seen others encounter, but I've yet to find a clear answer on how to solve from any of their threads.

I've tried installing amd64 12.04 Server and amd64 12.10 Server on a Supermicro SuperServer 6015B-NT and had zero luck. Install media = usb stick. This is a dual proc Xeon E5420 2.5GHz machine w/28GB of RAM and an Adaptec 5405 RAID controller, with 4 x 3TB Toshiba 7200 rpm SATA drives configured for RAID 10 (~6TB array).

Problem:

- Boot fails, usually somewhere after a statement like this:

[t1] EXT4-fs (dm-0): INFO: recovery required on readonly filesystem
[t2] EXT4-fs (dm-0): write access will be enabled during recovery
[t3] EXT4-fs (dm-0): recovery complete
[t4] EXT4-fs (dm-0): mounted filesystem with ordered data mode. Opt: (null)
Begin: Running /scripts/local-bottom ... done.
done.
Begin: Running /scripts/init-bottom ... done.
[t5] Adding 29356028k swap on /dev/mapper/host-swap_1. Priority:-1 extents:1 across:29356028k
[t6] EXT4-fs (dm-0): re-mounted. Opts: errors=remount-ro

and then... haaaaang (ctrl-alt-del doesn't work)

- Recovery mode also fails to boot, usually after what appears to be an ATI Radeon driver issue or an ACPI error:

[t1] ACPI Warning: 0x00000000fed1f410-0x00000000fed1f414 SystemMemory conflicts with Region \RCRB 1 (20120320/utaddress-251)
[t2] ACPI: If an ACPI driver is available for this device, you should use it instead of the native driver
[t3] lpc_ich: Resource conflict(s) found affecting iTCO_wdt
[t4] EDAC MC: Ver: 2.1.0
[t5] [drm] Initialized drm 1.1.0 20060810
[t6] gpio_ich: ACPI BAR is busy, GPI 0 - 15 unavailable
[t7] gpio_ich: GPIO from 206 to 255 on gpio_ich
[t8] dca service started, version 1.12.1

and then.... haaaaaang (ctrl-alt-del doesn't work)

Radeon break looks a little more like this:

[t3] lpc_ich: Resource conflict(s) found affecting iTCO_wdt
[t4] spchp: Standard Hot Plug PCI Controller Driver version: 0.4
[t5] gpio_ich: ACPI BAR is busy, GPI 0 - 15 unavailable
[t6] gpio_ich: GPIO from 206 to 255 on gpio_ich
[t7] ioatdma: Intel(R) QuickData Technology Driver 4.00
[t8] EDAC MC: Ver: 2.1.0
[t9] [drm] Initialized drm 1.1.0 20060810
[t10] [drm] VGACON disable radeon kernel modesetting
[t11] [drm] Support vblank timestamp caching Rev 1 (10.10.2010).
[t12] [drm] No driver support for vblank timestamp query.
[t13] [drm] Intialized radeon 1.33.0 20080528 for 0000:09:01.0 on minor 0

and then.... haaaaaang (ctrl-alt-del doesn't work)

- Rescue Mode provides a little bit of detail:
I've typically elected to Execute a shell in /dev/host/root:

# cat /etc/fstab

# /etc/fstab: static file system information
# warning to use 'blkid'
# <file system> <mount point> <type> <options> <dump> <pass>
/dev/mapper/host-root  / ext4 errors=remount-ro 0 1
# /boot was on /dev/sda2 during installation
UUID=158...-...ebe  /boot ext2 defaults 0 2
/dev/mapper/host-swap_1  none swap sw 0 0
/dev/fd0  /media/floppy0 auto rw,user,noauto,exec,utf8  0  0


# blkid
/dev/sda2:  UUID="158...-...ebe" TYPE="ext2"
/dev/sda3:  UUID="fhh...-...f01"  TYPE="LVM2_member:
/dev/mapper/host-root:  UUID="bd5...-...839"  TYPE="ext4"
/dev/mapper/host-swap_1:  UUID="a89...-...93d"  TYPE="swap"


# fdisk -l /dev/sda

WARNING: GPT detected on '/dev/sda/'! The util fdisk doesn't support GPT. Use gnu parted.

Disk /dev/sda: lots of info...
Units = secotrs of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000

Device Boot    Start     End     Blocks     Id   System
/dev/sda1       1         lots      lots        ee     GPT


# parted /dev/sda print
Model: Adaptec hostraid (scsi)
Disk  /dev/sda: 5991GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt

Number    Start      End        Size        File system     Name     Flags
 1             1049kB 2097kB  1049kB                                       bios_grub
 2             2097kB  258MB   256MB    ext2
 3             258MB    5991GB  5991GB                                     lvm


# cat /var/log/boot
(Nothing has been logged yet.)

# cat /var/log/dmesg
(Nothing has been logged yet.)


# cat /var/log/bootstrap.log

...
...
Processing triggers for initramfs-tools ...
#

Any help or suggestions are MOST appreciated! Thanks!

[jeffmefun]

Also... I've tried installing some things to see if they would help fix the problem, but I have no DNS support, so I'm having a hard time hitting any repositories, etc. ping 4.2.2.2 works, but ping www.google.com does not (unknown host).

Thanks!

[CharlesA]

Have you installed on this hardware before? If not, you could try popping in a single drive that isn't in the array and seeing it the install takes.

[MAFoElffen]

I think first we should first ask what this is on hardware wise...

- Branded PC? If so, what make/model?
- CPU?
- Is this a real serverboard or desktop mobo? Noticed the radion message / standard server graphics.
- What kind of drives and how many? (PATA/SATA/SCSI)
- Is the drive controller onboard or on the bus?
- What is your logical layout? Meaning is root on RAID with all installed under root or do you have things split out?
- What RAID type? Is the software RAID consisting of drives or mulitple smaller than drive partitions on drives?

I think the info from those questions might give a picture of what may be going on... Especially with the ACPI errors.

Next would be if this box can boot any version of any branch of Ubuntu's LiveCD's. Such as 10.04LTS, before KMS was implemented. That would help with trying to query the hardware and in diagnostics.

Even though it had md0 errors, I suspect from the other errors that it's having problems with hardware modules and drivers.

EDIT-- If it can't boot a LiveCD, no worries. There's still another way from the server install disk you used.

[jeffmefun]

Have you installed on this hardware before? If not, you could try popping in a single drive that isn't in the array and seeing it the install takes.

Hi Charles - yes, I actually had an install working (should have mentioned that) and managed to run some phoronix benchmarks, but after one reboot, it mysteriously stopped working.

Good idea about the array. I'll give that a shot with a drive out of the RAID controller.

[jeffmefun]

I think first we should first ask what this is on hardware wise...

- Branded PC? If so, what make/model?
- CPU?
- Is this a real serverboard or desktop mobo? Noticed the radion message / standard server graphics.
- What kind of drives and how many? (PATA/SATA/SCSI)
- Is the drive controller onboard or on the bus?
- What is your logical layout? Meaning is root on RAID with all installed under root or do you have things split out?
- What RAID type? Is the software RAID consisting of drives or mulitple smaller than drive partitions on drives?

I think the info from those questions might give a picture of what may be going on... Especially with the ACPI errors.

Next would be if this box can boot any version of any branch of Ubuntu's LiveCD's. Such as 10.04LTS, before KMS was implemented. That would help with trying to query the hardware and in diagnostics.

Even though it had md0 errors, I suspect from the other errors that it's having problems with hardware modules and drivers.

EDIT-- If it can't boot a LiveCD, no worries. There's still another way from the server install disk you used.

Hi Mafo - thanks for the questions. I believe most of them are answered in the original post. I'm not sure it had any md0 errors, as there is no md device.

I can boot from a LiveCD, or at least, I can boot from the server install usb key.

Good suggestions about the 10.04LTS - I'll check that one out.

Thanks! Jeff

[CharlesA]

Try booting into an older kernel if you had a kernel update recently.

[jeffmefun]

Hi CharlesA - I wish I did! This is for a fresh install (I know... old machine, new OS!). Thanks! Jeff

[darkod]

It's probably too late now, but is there any reason you decided to get a HW raid card?

Many people, me included, would prefer ubuntu server running mdadm SW raid, but for that the disks have to be passed straight through, without any raid on the card. For that to work you need a simple HBA instead of raid card, something like the IBM M1015 flashed to IT mode. In that mode the card only passes the disks through and provides connections to all disks when there are no sufficient ports on the motherboard.
The Ubuntu OS is handling the SW raid with mdadm.

This is more flexible solution since it allows you array reshapes, and more important, is not HW dependent. If your HW raid controller dies, you need the same model to get your array going without data loss, and it's a question if it will work even then.

With mdadm raid you can connect the disks on any linux machine and the data is easily accessible.

[MAFoElffen]

Gosh. ::embarassed:: That was right before I went to sleep last night. Too many hours, not enough time. LOL. I guess I should have read that first line instead of focusing on the errors...

Non-intrusive- On any server install disk, on any server install... After the time server dialog... between that dialog and the partitoner, it does hardware probes. At the time of the partioner, it gives you the choice to go back, which will go to the install template menu, which the last option exists the install. If you exit at that time, there would be not changes to what is installed.

The server install diplays in tty1, whereas the sys messages display in tty4. After the time server dialog, switch over to tty4 and follow the sys messages... Look for the messages about when it hits that RAID controller and what driver it defaults to... Or fails to find and falls back to. Then when it hits the drives.

Before doing that, you might want to look at this:
http://manpages.ubuntu.com/manpages/....4freebsd.html

From a LiveCD, you might want to chroot the installed sys and check if that module is there...

On HW RAID... I use both hw and md... depending what is there. That Adaptec card is usually auto-detected with no problem.