Missing devicefiles for partitions

marjue_1 · n00b Joined: 06 Mar 2013 Posts: 4

Hi

I have a new installed Gentoo server with software raid on it. I developed this system on a temporary hardware and everything was fine.

Now after the move on the final hardware i have the following problem.

The system boots up fine.

The raid is damaged:

NeddySeagoon · Posted: Wed Mar 06, 2013 6:43 pm Post subject:

marjue_1,

Welcome to Gentoo.

You have a partition table on /dev/sdb, that does not mean there is anything behind it. Lets mount /dev/sdb1, in readonly mode as a test.

marjue_1 · n00b Joined: 06 Mar 2013 Posts: 4

Hi

I have moutet /dev/sdb as you suggested. I can see the files from the /boot filesystem which is normally mouted from /dev/md126

marjue_1 · n00b Joined: 06 Mar 2013 Posts: 4

Here the dmsg

there is a "[ 3.414138] sdb: unable to read partition table". But why! The drive is working. I have already re-written the table.
Is it possible to rescan the IDE-bus. I know a script from SuSE to rescan the SCSI-bus for new drives during uptime.
Because of the old hardware i was forced to plug the drives on a single parallel-IDE Port as master/slave (jumper is on drive-select).
In the developing machine the drives have been on two controllers as master. Is there something missing in the kernel?

The pastebin'ed dmsg

http://de.pastebin.ca/2329461

Bye

Marcus

NeddySeagoon · Posted: Thu Mar 07, 2013 9:15 pm Post subject:

marjue_1,

Jumpering the drives to cable select only works when
a) your motherboard supports cable select
b) you use 80 wire IDE cables
c) your drives support cable select.
d) the IDE cable is fitted the right way round - it has a motherboard end[1].

I suspect something does not support cable select.
The good news is that you can emulate it.

Set the jumpers on the drives as follows:-
Set the drive at the end of the IDE cable to master
Set the drive in the middle of the IDE cable to slave.
Thats what cable select does, so setting the jumpers this way is harmless, even if you do have working cable select.

The IDE bus cannot be rescanned. Unlike SCSI it does not support hot plugging devices, so there is no point.

You will hate raid on two drives on the same IDE cable as commands to the drives cannot be overlapped.
This makes it very slow.

dmesg shows that sdb appears to be present but that it cannot be read - see all the resets and errors for sdb.
This may be as a result of both drives attempting to respond to commands sent to sdb because your cable select doesn't work.

[1]to determine the IDE cable ends, look at the cable close to the two end connectors. Notice that at one end, one wire is cut short.
The end with one wire cut short goes to the master drive.
This wire is connected to 0v on the motherboard and is connected to the slave drive only, where the 0v sets the drive to slave mode.
As its not connected to the master, the voltage there remains high (5v), which selects master mode.
This only works with 80 wire IDE cables.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

marjue_1 · n00b Joined: 06 Mar 2013 Posts: 4

Hi NeddySeagoon

Yes i know all these limitations. I was forced to do so because i only had old hardware for the developing machine and the production machine is not really better :roll:

But i didn't know that cable select only works on 80-pin cables

I will set the jumpers to real master and slave and check the other things you told me

And then i will ask the boss of the company for SATA drives

Thanks for help .... i will report in some days

Bye
Marcus

_______0 · Guru Joined: 15 Oct 2012 Posts: 521

cheap ass boss