IDE hard drives not visible

This problem appears to be a kernel issue. When I select kernel 2.6.22-14-generic from the grub menu during boot up, the IDE addressing behaves normally, and I can access all my hard drives. When I select kernel 2.6.24-16-generic the IDE addressing is screwed, and I can only access some of my hard drives.

Note also that at no time was there a sdc or sdd. When running the 2.6.24-16-generic kernel, there is at most just sda and sdb. I have three hard drives, and with the 2.6.22-14-generic kernel there is hda hdb and hdc.

Just checking back through prior bug reports. I found two bug#222115 and bug#221677 that may relate to this one. Going further back there are several other complaints of not being able to access a CD drive.

When I attempted to install Hardy for the first time, the install location was on hdc the master on the secondary controller. The install completed, and upon reboot, it reported file not found. I believed at the time this was due to the grub loader getting confused about where things are located.

I swapped the IDE controller cables, to make the install location on hda the master on the primary controller. The install is now successful excepting that I can not address all my hard drives.

It makes no difference if Hardy is installed from disk, or upgraded from a prior version. Previously I reported that the problem happens with the 2.6.24-16-generic kernel, and not with the 2.6.22-14-generic kernel. This was done with all else being the same. That is, I booted into the same partition, only selecting a different kernel in the grub boot menu.

Attached is the output of lshw taken when running Dapper from the second partition on the master on the primary IDE controller.

Attached is the output of lshw taken when running Hardy from the first partition on the master on the primary IDE controller.

Looks like bug#93655. Is your /etc/fstab migrated to uuid? If it's not, you should run /usr/lib/udev/migrate-fstab-to-uuid.sh

No Darko, it is not to do with uuid in the fstab.

For the record, the fstab does use UUID to identify the disks.

Please read more carefully. You will see that I have stated, that fdisk -l does not report all my disks. Furhermore you will see that I have included output from lshw. If you would kindly take the time to examine the output of lshw from Dapper and Hardy installs, you will see that a disk is missing from the Hardy install that is present in the Dapper install. Note that this output was taken from my current hardware configuration. As I have stated above, Dapper is installed in the second partition on the master of the primary controller. Hardy is installed on the first partition on the master of the primary controller.

Furthermore please note that when attempting to install Hardy in a disk on the seconday IDE controller, the install fails because it fails in the kernel to identify all my drives. I suspect the reason for the file not found message I stated above, was that it installed on the disk on the secondary controller, but then on reboot, the disk was no longer visible.

It does not appear to me to be a configuration issue, but rather a kernel issue. Why? As I stated above, the problem also exists in Fedora Core 8, also by booting Hardy but with the Gutsy kernel, all disks are identified properly. Simply changing the kernel and only the kernel to an earlier kernel, the system is fully operational with all disks correctly identified in fdisk -l.

Point taken regarding use of sda in place of hda. I was not aware of this change. It would seem there are bugs in this change over.

Configurations that failed for me in Hardy:

1) Two drives, with one as master on the primary controller, and one as master on the secondary controller. Only one drive shows up in fdisk -l

2) Three drives, with two on the primary controller and the other as master on the secondary. Only two drives show up.

3) Three drives, with two on the secondary controller and the other as master on the primary. Only two drives show up.

At the moment, I am running Hardy with two drives on the primary controller, with both drives accessible, and the third as master on the secondary controller, and is not visible in fdisk -l.

One other thing I will mention. When running the live Hardy CD, the same symptoms exist. I can see two of my three drives in fdisk -l. Definitely need a new kernel.

I think I have the same problem as you guys, after gutsy->hardy upgrade I lost my cdrom drive which used to be addressed as /dev/hda (my HD is serial ata and so not connected to the standard IDE bus). I only have a /dev/sr0 and /dev/scd0 entries which relate to scsi emulation stuff, I think. The cdrom is in fact undetected by all the ubuntu high-level subsystem (gnome, and stuff). I tried to do a fresh install and the installer stops telling me it's unable to mount the cdrom drive.

I think a patch to the kernel to fix this is quite urgent because it makes many systems almost unusable.

I posted this problem on Bugzilla. They have a kernel update available that I have been asked to try. I will try with Fedora Core 8 later Today when I can get some time to test this.

They also asked for the output of dmesg. So I have attached the output of dmesg here for my Hardy install.

I booted into Hardy, opened a terminal window, then as root run fdisk -l, and captured the output of dmesg to file as attached.

I add a little thing about my fresh install try: before installation hung telling not being able to mount the cdrom drive, the screen was flooded but such messages (process stalled for a couple of minutes):

ata3.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
ata3.00: cmd a0/01:00:00:20:00/00:00:00:00:00/a0 tag 0 dma 32 in
         cdb 46 00 00 00 00 00 00 00 20 00 00 00 00 00 00 00
         res 40/00:02:00:0c:00/00:00:00:00:00/a0 Emask 0x4 (timeout)
ata3.00: status: { DRDY }
ata3: soft resetting link
ata3.00: configured for UDMA/25
ata3: EH complete

I also attach my box fstab, if it can be in any way useful.

errata: I meant dmesg, not fstab ;^)

i have the same problem here, complete new Hardy Server installation
my second harddrive is unmountable, same error messages than scananza.
I am using the gutsy kernel now on the machine (some other things are not working like apt-cdrom add), however i will wait, until there is a kernel-patch available, before i start using the machine properly

I am now thinking this problem may be related to the specific hardware that I am using. The full hardware listing is included above in the output of lshw when run on Dapper. The one from Hardy is missing one drive.

There are a couple of warning messages in the dmesg listing included above.
Driver 'sd' needs updating - please use bus_type methods
Driver 'sr' needs updating - please use bus_type methods
Some people suggest that these messages are of no consequence. I don't know, it just what I have read.

I think this problem is in the kernel, and will require a new kernel to fix this.

Thanks to the Red Hat team, I am now able to access all my hard drives. Refer to https://bugzilla.redhat.com/show_bug.cgi?id=444334 for details.

The problem for me turns out to be an issue with how the IDE cable is connected. The center connector is near one end, and the end furthest away from the center connector is typically connected to the controller on the motherboard. In my case, the end typically connected to the motherboard was connected to the CD drive, and the center connector to the controller on the motherboard, with the short end going to the hard drive.

I relocated the hard drive to be near the CD drive, and using the two connectors near each other, connected the CD drive, and to the hard drive, with the long end going to the controller on the motherboard.

Other than relocate the hard drive, the only change I made is to the IDE cabling. Now it works fine. I can access all three hard drives.

One last comment. For the less technically competent user than I, relocating drives and changing cabling, would be prohibitive. The average home user may not be able to fix this problem, and for some users, they would be unable to install Hardy. I recommend that this bug report remain open until a fix is made to the kernel/libata that is more tolerant of cabling.

Hi Don,
If found out that I also had the longest tail of the IDE cable attached to the CDROM and not to the MB but, unfortunately, that didn't solve the problem (I also switched to a brand new IDE cable, just to be sure, since the old one was pretty twisted and deformed).

My setup is quite simple: just the CDROM as primary master, there are no other devices, the second bus is empty.
I tried to connect the CDROM both to the middle and to the end connector but no way.
BIOS recongnize correctly the device as primary master and the kernel detects a scsi-emulated cdrom, loads the proper devices /dev/sr0 and /dev/scd0 but NOT the standard /dev/hda. For the rest of the operating system the cdrom simply doesn't exist and forceing apps to use /dev/sr0 just hangs them.

This bug is really annoying, if it's kernel-related stuff, I hope a new patched kernel will be added to repositories as soon as possible.

Scananza,

Follow the link above to the Red Had Bugzilla site for the full resolution discussion. There it is stated, that problems exist with both cabling and jumper settings. In my case, recabling with the long end to the controller, fixed my problem. The bios handles the boot hugely faster now. My jumper settings are for the hard drive as master with slave present, and for the CD/DVD as slave.

In your case, make sure the long end of the cable is connected to the controller, and that the jumper settings on the back of the CD drive are configured for master. Not cable select or anything else.

I've checked out the jumper, it's correctly closed on MAster; the cable has the long edge connector attached to the MB and the opposite side is stuck into the CDROM, the central connector is free.

For the sake of (computer) science, I can try if setting it to SLave or CableSelect makes it work... :^D

Later on today I will compile kernel 2.6.25 myself and see if the problem persists even with the latest IDE driver.

....just for fun I've just effectively tried it SLave, no way (I would have been surprised of the contrary, actually)

Tried 2.6.25 kernel and problem went away, so for me it's totally a kernel issue. Hope a fix will be released soon for 2.6.24 serie since nvidia subsystem won't compile against my 2.6.25 vanilla sources... :^\

I added the bug watch by mistake. Try to revert...

I had this same problem (IDE HDD and DVD-ROM wouldn't show up automatically in BIOS, and wouldn't show up in Hardy even when I got the BIOS to recognize them): I re-cabled with the long end to the controller and it fixed the problem.

THANKS!

Funny enough, my problem went away as well. My solution was kind of crazy, but it worked. I have a master/drive slave/cdrom and another master. Problems always came with the second master.

After rejumpering the first master to cableselect (no master!) AND removing the cdromslave, my second harddrive is suddenly detected and works. Don't ask, i have no clue why, it makes absolutely no sense at all, but it works :-)

Per Don's comment - https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/222322/comments/14 - I'm going to close this report. For anyone still experiencing issues, the Ubuntu kernel team is actively developing the upcoming Intrepid Ibex 8.10 release. If you could test the latest Alpha for Intrepid - http://www.ubuntu.com/testing and open a new report if your issue still exists that would be great. Thanks.