PDA

View Full Version : [ubuntu] 12.04 + fakeraid + pysdm



Tranas
November 3rd, 2012, 05:20 PM
Have a functional dual boot 12.04/XP with fakeraid on an external card.

After log on in 12.04 the fakeraid is visible, can be mounted manually, the array is seen by 12.04 as a single drive and 12.04 has no issues reading and writing.

pysdm was installed in an effort to mount the fakeraid automatically, however pysdm sees the array as two separate drives and cannot configure them.

Is there a way to mount the raid automatically by editing the fstab file manually? - and since pysdm appears to be a graphical interface for doing the exact same thing [but fails] do I have an underlying problem that needs to be addressed?

sudo fdisk -l follows

TIA


Disk /dev/sda: 250.1 GB, 250059350016 bytes
255 heads, 63 sectors/track, 30401 cylinders, total 488397168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x0003c255

Device Boot Start End Blocks Id System
/dev/sda1 * 63 20482874 10241406 7 HPFS/NTFS/exFAT
/dev/sda2 20482875 464007167 221762146+ 7 HPFS/NTFS/exFAT
/dev/sda3 464007168 484487167 10240000 83 Linux
/dev/sda4 484489214 488394751 1952769 5 Extended
/dev/sda5 484489216 488394751 1952768 82 Linux swap / Solaris

Disk /dev/sdb: 20.5 GB, 20520493056 bytes
255 heads, 63 sectors/track, 2494 cylinders, total 40079088 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x963aedd5

Device Boot Start End Blocks Id System
/dev/sdb1 * 2048 35155967 17576960 83 Linux
/dev/sdb2 35158014 40077311 2459649 5 Extended
/dev/sdb5 35158016 40077311 2459648 82 Linux swap / Solaris

Disk /dev/sdc: 1500.3 GB, 1500301910016 bytes
255 heads, 63 sectors/track, 182401 cylinders, total 2930277168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x15af43bf

Device Boot Start End Blocks Id System
/dev/sdc1 63 2929677659 1464838798+ 7 HPFS/NTFS/exFAT
Partition 1 does not start on physical sector boundary.

Disk /dev/sdd: 1500.3 GB, 1500301910016 bytes
255 heads, 63 sectors/track, 182401 cylinders, total 2930277168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x15af43bf

Device Boot Start End Blocks Id System
/dev/sdd1 63 2929677659 1464838798+ 7 HPFS/NTFS/exFAT
Partition 1 does not start on physical sector boundary.

Disk /dev/mapper/pdc_ddbhgeedgg: 1500.0 GB, 1499999961088 bytes
255 heads, 63 sectors/track, 182364 cylinders, total 2929687424 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x15af43bf

Device Boot Start End Blocks Id System
/dev/mapper/pdc_ddbhgeedgg1 63 2929677659 1464838798+ 7 HPFS/NTFS/exFAT
Partition 1 does not start on physical sector boundary.

Disk /dev/mapper/pdc_ddbhgeedgg1: 1500.0 GB, 1499994929664 bytes
255 heads, 63 sectors/track, 182363 cylinders, total 2929677597 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Alignment offset: 512 bytes
Disk identifier: 0x69205244

This doesn't look like a partition table
Probably you selected the wrong device.

Device Boot Start End Blocks Id System
/dev/mapper/pdc_ddbhgeedgg1p1 ? 218129509 1920119918 850995205 72 Unknown
Partition 1 does not start on physical sector boundary.
/dev/mapper/pdc_ddbhgeedgg1p2 ? 729050177 1273024900 271987362 74 Unknown
/dev/mapper/pdc_ddbhgeedgg1p3 ? 168653938 168653938 0 65 Novell Netware 386
Partition 3 does not start on physical sector boundary.
/dev/mapper/pdc_ddbhgeedgg1p4 2692939776 2692991410 25817+ 0 Empty
Partition 4 does not start on physical sector boundary.

Partition table entries are not in disk order

darkod
November 3rd, 2012, 05:47 PM
Hmm, that fakeraid looks very weird. It looks like there are two raid1 arrays one into the other. The partitions that are relevant look like the ones ending in ...gg1p1, ...gg1p2 and ...gg1p3.

So, depending which one you want to mount, and whether the filesystem is ntfs or different, you would need to add a line in /etc/fstab something like:

/dev/mapper/...gg1pX /mount-point ntfs defaults 0 0

That should do it.

Tranas
November 3rd, 2012, 06:29 PM
Hmm, that fakeraid looks very weird. It looks like there are two raid1 arrays one into the other. The partitions that are relevant look like the ones ending in ...gg1p1, ...gg1p2 and ...gg1p3.

So, depending which one you want to mount, and whether the filesystem is ntfs or different, you would need to add a line in /etc/fstab something like:

/dev/mapper/...gg1pX /mount-point ntfs defaults 0 0

That should do it.

Thanks Darko

Does this shed any light on it?

sudo dmraid -ay
RAID set "pdc_ddbhgeedgg" already active
RAID set "pdc_ddbhgeedgg1" already active

I gather that to be sbc and sbd [which are not visible separately in nautilus]. Nautilus displays the same volume name as Windows.


from: help*Ubuntu*com

You need to know the name of your raid set

sudo dmraid -ay

RAID set "pdc_ddbhgeedgg" already active
RAID set "pdc_ddbhgeedgg1" already active

Edit fstab as per instructions above :

nano -w /etc/fstab
An example line to add

/dev/mapper/pdc_ddbhgeedgg1 /media/raid reiserfs user,nosuid,exec,nodev 0 0

Make sure you create the directory /media/raid

mkdir /media/raid
Reboot


Significant differences....
Does "pdc_ddbhgeedgg1" represent the array or an set member?
my fear is that editing fstab could break the array by mounting "half" of the array, as pysdm displays the array in a totally bogus manner.

pysdm sees sdc and sdd as partitions in sda.

TIA

darkod
November 3rd, 2012, 06:37 PM
That's why I said it looks weird. They are both raid sets (not raid member disks), with and without the 1 at the end.

Usually you have a single raid set named /dev/mapper/blah_blah and then the partitions on it end in p1, p2, etc.

You have raid set, then another raid set inside it and ending with the 1, and then the partitions on it gg1p1, gg1p2, etc.

I don't know what to recommend. There shouldn't be two sets when there is only one raid1 device.

Trying to mount it shouldn't break it in any case since you are not trying to use the separate disks /dev/sdc and /dev/sdd. You are trying to use the partitions on the raid set, and since the partitions are ...gg1p1, ...gg1p2 that's why I suggested to try with that first.

Of course, the mount point (folder) needs to be created first and existing.

Tranas
November 3rd, 2012, 06:38 PM
Hmm, that fakeraid looks very weird.

screenshot of psydm

darkod
November 3rd, 2012, 07:31 PM
I don't know how that program works.

I would use terminal and the command line, edit /etc/fstab as mentioned earlier and see if it works.

fdisk in terminal obviously detects the raid set, the only thing is that there seems to be two of them when we are expecting one.

Tranas
November 3rd, 2012, 08:16 PM
I don't know how that program works.

I would use terminal and the command line, edit /etc/fstab as mentioned earlier and see if it works.

fdisk in terminal obviously detects the raid set, the only thing is that there seems to be two of them when we are expecting one.


fyi -

PySDM is a Storage Device Manager that allows full customization of hard disk mountpoints without manually access to fstab.
It also allows the creation of udev rules for dynamic configuration of storage devices

Features
PyGTK graphical interface
Management of fstab file
Partition autodetection
Automated configuration for new devices
Filesystem-dependent options. ext2, ext3, fat, ntfs, reiserfs, swap and xfs supported now, more coming soon... :)
Mounting management
Hotpluging
udev rules management
Device naming
Permission assignement
internationalization and localization support

my understanding, it is a graphical interface for manipulating fstab [we use on other machines to automatically mount drives]

I am going to pursue a little more information before going forward.

As you can see in the posted png, sda2 presents as sdc. pysdm also sees sda3 as sdd and *both* have the same volume name. In essence it sees the physical raid card with both array drives separately within sda [which is not a raid set]. I can theoretically set pysdm to mount sda2 or sda3 - just not willing to risk it.

Since 12.04 as able to mount the array via Nautilus without issues [which could just be dumb luck ;-) ], pysdm realistically should be able to do the same.

Appreciate the help Darko - will post when there more clarity

Tranas
November 8th, 2012, 06:25 AM
That's why I said it looks weird. They are both raid sets (not raid member disks), with and without the 1 at the end.

There shouldn't be two sets when there is only one raid1 device.



There is something basically bogus in the way some other GUI apps deal with fakeraid which cascades to other partitions on the local machine - specifically both system-config-samba and pysdm. I presume that the output of fdisk -l is generated by the linux kernel, not by 12.04

Quite obviously, 12.04 sees and mounts the devices/partitions manually. With Samba installed, Windows machines on the local network can access and map these manually mounted drives R/W without issues.

Samba, however, cannot deal with the drives/partitions on the local machine nor can other Samba servers on the local network. Samba is unable to mount the partitions locally or over the network. On the local machine these same mounted partitions are accessible via Nautilus through devices or /media without issue.

Samba, however, only sees the machine name and the partition names but cannot "mount" them - evidently the internal mapping is a mystery to Samba, just like it is to psydm. I presume that both are looking for a partition name like sdx1 which in fact refers to the raid array and instead get the convoluted mapping we see in the fdisk result. Interesting to note that both clonezilla and gparted seem to see the partition structure properly.

My guess at this point is that both system-config-samba nor pysdm are unable to properly deal with the partition structure. One would conclude that whatever code 12.04 uses to translate that structure needs to be added to both system-config-samba and pysdm.

ymmv

darkod
November 8th, 2012, 09:22 AM
As I said, I have no idea about pysdm. I understand they say it's only a front for another app, but don't forget that fronts not always work as expected at the baseline. That's why webmin is not recommended and is not in the repositories for example.

As for Samba, I would expect it can work and recognize fakeraid without issues. The question is are you doing it in the terminal or using some front also?

In terminal if you open /etc/samba/smb.conf and configure the shares there I don't think it should have issues. I might be wrong though.

Just about a week ago I reinstalled my home network storage after buying two new 2TB disks. They are set up in raid1 with mdadm (software raid). Samba is simply using the /mount-point you specify for a share, and that's it. It shouldn't care if the system is on top of mdadm raid, or fakeraid, or single disk... That's in theory at least.

Tranas
November 8th, 2012, 09:53 AM
As I said, I have no idea about pysdm. I understand they say it's only a front for another app, but don't forget that fronts not always work as expected at the baseline. That's why webmin is not recommended and is not in the repositories for example.

As for Samba, I would expect it can work and recognize fakeraid without issues. The question is are you doing it in the terminal or using some front also?

In terminal if you open /etc/samba/smb.conf and configure the shares there I don't think it should have issues. I might be wrong though.

Just about a week ago I reinstalled my home network storage after buying two new 2TB disks. They are set up in raid1 with mdadm (software raid). Samba is simply using the /mount-point you specify for a share, and that's it. It shouldn't care if the system is on top of mdadm raid, or fakeraid, or single disk... That's in theory at least.

Both of the proggies mentioned are GUI front ends. For those of us who are command-line challenged, they are usually pretty slick. This is obviously an exception.

this is the piece of the smb.conf that is likely relevant, and I presume that uncommenting the browseable entry might solve the Samba part of the issue?

[Data2]
path = /media/Data2
writeable = yes
; browseable = yes
guest ok = yes
comment = 1.5 Tib Raid Array

[Data]
path = /media/Data
writeable = yes
; browseable = yes
guest ok = yes
comment = sdb2 on BX5-U

The GUI only offers "Writeable" and "Visible" as configurable options - presumably, checking the "Visible" option should have edited the smb.config to uncomment browseable.

darkod
November 8th, 2012, 10:00 AM
If you want an open for all share, without any permissions restrictions, look into the smb.conf in the [global] section at the security option. For open shares it should be:
security = share

Then in the share definitions you posted, my open shares have only:

[share name]
comment = any comment
path = /path/to/folder
guest ok = yes
read only = no

That allows it to accept guests (no login details needed), and the read only = no makes it writeable. I don't use the browseable option and it works. But i also have security = share in [global] as mentioned.

You can stick with the current share definitions, uncomment browseable first and restart the samba service:
sudo service smbd restart

Check if that changed anything.

If it still doesn't work, you can consider making a copy of the smb.conf so you can keep the current content, and then edit the share definitions similar to my shares.

Tranas
November 8th, 2012, 10:04 AM
clip of the Nautilus screen attached

darkod
November 8th, 2012, 10:40 AM
I don't understand it, is this on another computer?

The window on the right which is only partly shown has the share Data mounted and displays the content.

Tranas
November 8th, 2012, 11:27 AM
I don't understand it, is this on another computer?

The window on the right which is only partly shown has the share Data mounted and displays the content.

lol

yep, same machine - the lower right is Nautilus opened on the same machine, simply with a new window.

As posted earlier - 12.04 (and windows machines on the network) can mount browse and R/W, but Samba on the same machine or Samba servers on the network cannot.

BTW uncommenting the earlier lines did not work.

darkod
November 8th, 2012, 11:45 AM
I'm lost. Why would you try opening it with Samba or Samba servers? The server is to serve, provide the service, and it seems it's doing that since clients can access the shares.

It's not to try and open the same share from Samba. First of all, I'm not sure there would be a conflict since the share is mounted already on the same machine at /media/Data.

If the server provides the service and the clients can connect to it and use it, what more do you want?

Tranas
November 8th, 2012, 12:24 PM
I'm lost. Why would you try opening it with Samba or Samba servers? The server is to serve, provide the service, and it seems it's doing that since clients can access the shares.

It's not to try and open the same share from Samba. First of all, I'm not sure there would be a conflict since the share is mounted already on the same machine at /media/Data.

If the server provides the service and the clients can connect to it and use it, what more do you want?

See attached clip from another 12.04 machine (BX6-U) on the same network. Obviously cannot access the partition "Data2" on BX5-U. The Server on BX5-U is not serving.

I repeat - Samba servers on the same network are unable to mount the partitions on BX5-U over the network. As you can see from the clip, BX6-U has no issues displaying or accessing it's own Samba shares.

I believe you are incorrect about What a Samba server can serve up. Why you would "want" to open 1 or 10 new windows is not the issue. Simple fact is, in Nautilus, BX5-U should be able to browse it's own partition "Data2" in Devices, in File system/Media or in Network using as many windows as you choose to open. No?

darkod
November 8th, 2012, 12:33 PM
It's easy for you to speak, you know your setup. I don't. :)

Even with the two screenshots it's not easy to figure out which server is which and what is exactly hapenning.

One thing I see that could potentially be an issue, is that it seems you are using exactly the same path on both servers, /media/Data.

On top of that, by simply clicking the share in Nautilus it will try to mount it in /media, so there might be a clash with the already existing (local) /media/Data.

On a server that CAN NOT mount the BX5 share over the network, in terminal try this:

sudo mount -t cifs //<BX5 IP>/Data /mnt -o guest
ls /mnt

Can that mount the Data share to /mnt and list its content?

If it can, obviously one server can access the other one. The equal mount point and using /media might be the issue.

Tranas
November 8th, 2012, 12:49 PM
It's easy for you to speak, you know your setup. I don't. :)

Even with the two screenshots it's not easy to figure out which server is which and what is exactly hapenning.



I understand the difficulty and appreciate your patience.

From the attached clip you can see that the same server on BX5-U allows an XP machine on the same network to browse files.

I do not believe the fact that both are /media/data2 should be an issue since both Samba servers know their own names - obviously dispayed in the menu bar of the XP machine.

I will try the code...

darkod
November 8th, 2012, 12:55 PM
I do not believe the fact that both are /media/data2 should be an issue since both Samba servers know their own names - obviously dispayed in the menu bar of the XP machine.

I will try the code...

Correct, but the machine name is not the mount point. You got me thinking now.

When you have a usb stick and the partition on it has a label USB, it auto mounts in /media/USB.

I don't know where samba shares mount, never looked into it really, but if server1 tries to mount the Data share of server2 also automatically in /media/Data, it will clash with /media/Data that server1 already has.

I suggest not to use /media for the samba shares, but lets see first if mounting on a specific mount point works. Later you can consider if you want to copy the data from /media/Data and /media/Data1 to another location, if you have the free space and before you have filled up the server. /media is sort of reserved for devices outside the filesystem, that's why they show up under Devices in Nautilus.

You can create any mount point that you want inside /, the only difference will be you will have to reach it through Filesystem in Nautilus (in the section Computer). But it will be much cleaner setup since the local data is outside /media, as it should be.

Tranas
November 8th, 2012, 01:02 PM
Correct, but the machine name is not the mount point. You got me thinking now.

When you have a usb stick and the partition on it has a label USB, it auto mounts in /media/USB.

I don't know where samba shares mount, never looked into it really, but if server1 tries to mount the Data share of server2 also automatically in /media/Data, it will clash with /media/Data that server1 already has.



Simple way of checking that theory is shut down BX6-U to eliminate all possibility of conflict in naming, reboot BX5-U and see what shakes. Back shortly...

darkod
November 8th, 2012, 01:07 PM
Simple way of checking that theory is shut down BX6-U to eliminate all possibility of conflict in naming, reboot BX5-U and see what shakes. Back shortly...

It's not exactly the same. The issue is mounting the samba share of one server to the other, right? So you need both of them running.

If you shut down one, you only have the local /media/Data from the running server and you can't try mounting the Data share of the first server since it's off.

Tranas
November 8th, 2012, 01:30 PM
It's not exactly the same. The issue is mounting the samba share of one server to the other, right? So you need both of them running.

If you shut down one, you only have the local /media/Data from the running server and you can't try mounting the Data share of the first server since it's off.

Well, sort of.

BX5-U has no issues mounting the shares on BX6-U. The issue is that it is not serving up it's own shares either to other Samba servers or to itself.

By the way, the same machine, when booted to Windows on the same network allows all machines, including Samba servers to browse that very same fakeraid array. There are no issues.

Your point that the machine name is not the mount point is in my view correct - therein lies the rub.

Samba (on any machine) is not able to translate the mapping for BX5-U visible in fdisk -l in a way that enables Samba to "find" Data2 on 'BX5-U server. The code for the GUI in 12.04 however, does not have that problem. It is able to do everything it needs to do with Data2. Only when Samba gets in between the two is there an issue.

Another point is that there is NO fakeraid array on BX6-U. The translation of the fakeraid array mapping is what I believe to be the issue for the Samba accessing shares on BX5-U, either locally or over the network.

Tranas
November 8th, 2012, 02:01 PM
On a server that CAN NOT mount the BX5 share over the network, in terminal try this:

sudo mount -t cifs //<BX5 IP>/Data /mnt -o guest
ls /mnt

Can that mount the Data share to /mnt and list its content?



as follows if I understood the line correctly:

sudo mount -t cifs //192.168.1.179/Data /mnt -o guest ls /mnt
Usage: mount -V : print version
mount -h : print this help
mount : list mounted filesystems
mount -l : idem, including volume labels
So far the informational part. Next the mounting.
The command is `mount [-t fstype] something somewhere'.
Details found in /etc/fstab may be omitted.
mount -a [-t|-O] ... : mount all stuff from /etc/fstab
mount device : mount device at the known place
mount directory : mount known device here
mount -t type dev dir : ordinary mount command
Note that one does not really mount a device, one mounts
a filesystem (of the given type) found on the device.

... continues to nag with instructions...

*OR*

sudo mount -t cifs //192.168.1.179/Data /mnt -o guest
mount: wrong fs type, bad option, bad superblock on //192.168.1.179/Data,
missing codepage or helper program, or other error
(for several filesystems (e.g. nfs, cifs) you might
need a /sbin/mount.<type> helper program)
In some cases useful info is found in syslog - try
dmesg | tail or so


It does not, however, appear to mount "Data" - Nautilus continues to show the mount failure message.

darkod
November 8th, 2012, 02:22 PM
They were two different commands, like I wrote them. So, your second attempt was correct, but the message about wrong fs type or bad superblock might give us a hint. Maybe the filesystem has some corruption and it's refusing to mount the folder (share) properly.

I would boot the machine with the ubuntu standard cd in live mode, and run a fsck on the root partition (or on the partition where the share is, if different). It would be like:
sudo fsck /dev/sdaX

Note that the partition MUST NOT be mounted. That's why you need to do it from live mode.

Tranas
November 9th, 2012, 05:58 PM
They were two different commands, like I wrote them. So, your second attempt was correct, but the message about wrong fs type or bad superblock might give us a hint. Maybe the filesystem has some corruption and it's refusing to mount the folder (share) properly.

I would boot the machine with the ubuntu standard cd in live mode, and run a fsck on the root partition (or on the partition where the share is, if different). It would be like:
sudo fsck /dev/sdaX

Note that the partition MUST NOT be mounted. That's why you need to do it from live mode.

sdb1 is the Linux boot partition

ubuntu@ubuntu:~$ sudo fsck /dev/sdb1
fsck from util-linux 2.20.1
e2fsck 1.42 (29-Nov-2011)
/dev/sdb1: clean, 194831/1099440 files, 1079445/4394240 blocks
ubuntu@ubuntu:~$

also did a checkdisk on ntfs Data and Data2 from windows - both are clean

and the result then is

sudo mount -t cifs //192.168.1.179/Data /mnt -o guest

mount: wrong fs type, bad option, bad superblock on //192.168.1.179/Data,
missing codepage or helper program, or other error
(for several filesystems (e.g. nfs, cifs) you might
need a /sbin/mount.<type> helper program)
In some cases useful info is found in syslog - try
dmesg | tail or so

darkod
November 9th, 2012, 08:29 PM
Hold on, Data and Data2 are ntfs? I was under the impression they are folders inside your ubuntu server, which would mean ext3/ext4 or similar.

Tranas
November 9th, 2012, 09:29 PM
Hold on, Data and Data2 are ntfs? I was under the impression they are folders inside your ubuntu server, which would mean ext3/ext4 or similar.

Well, the fdisk -l is convoluted. Raid within raid that does not actually exist.

Ubuntu and/or Samba do not see them for what they are.
As you properly described - the layout described by fdisk -l is weird.

In the real world [that Gparted, Clonezilla and Windows see]:

Sda is is physical drive with XP [sda1] and an ntfs partiton [sda2] Data + partitions Sda 3 and 4 which were intended to be for Ubuntu + swap. [The last two are "there", but not used]

Data2 [sd-who knows what - c & d ?] is an ntfs raid array on a pci promise TX2300 card. Single partition. The matryoshka is visible via fdisk -l, weird or not. 12.04 sees the array as Data2 and deals with it normally. Mounts manually, reads and writes without trashing the array. The array presents normally in XP.

Sdb is a single physical drive, 2 partitions 1 for Ubuntu 12.04, 1 for swap. This is where the MBR on sda points to and multi-boots from.

So there are 3 physical drives.

12.04 seems to figure out how each is partitioned and mount/access as required [manually].

Samba is clueless.

The system dual boots.
12.04 can manually mount and browse everything, including other Samba servers on the network.
Samba cannot, even on the local machine.
Nothing can automount at bootup [so far].

darkod
November 9th, 2012, 09:51 PM
I already forgot about the raid weirdness. And I am not sure how samba would deal with ntfs. From what you said last, I understand that you are not auto mounting Data or Data2 in fstab, and even if you try it will not auto mount at boot.

When you put all of the above together, you have lots of issues. I'm not surprised it doesn't work.

First of all, is this to play around or a server that needs to be on most of the time? If it's a server that needs to be on, you can't seriously start talking about dual boot. Get rid of XP, and the ntfs partitions, and make everything linux. I think it will greatly improve the possibility of working properly.

If it's to play around, move the data temporarily and redo the raid since it definitely looks weird, something went wrong.

Tranas
November 9th, 2012, 11:02 PM
I already forgot about the raid weirdness. And I am not sure how samba would deal with ntfs. From what you said last, I understand that you are not auto mounting Data or Data2 in fstab, and even if you try it will not auto mount at boot.



Appreciate the help Darko - it has been enlightening as always.

Samba/Ubuntu obviously cannot deal with the fakeraid on the local machine. When the local machine is booted to windows, Samba has no problems whatsoever seeing the partitions over the network from other machines. Mounts and reads/writes to the very same raid drive - sees it as a single drive - which is the point of fakeraid.

Resolving Automount was the original intent of the thread -

"...Is there a way to mount the raid automatically "

which never got solved, because apparently, Samba/12.04 cannot deal with it.

Users of the local machine need to mount Data and Data2 manually using 12.04. PITA - as that is something that should be *simple*. [sort of like numlock on boot - but that is a whole other issue..]

If past is prologue, will just have to wait until the whoever deals with Samba solves the bug, because it eventually effects *them*.

This is not a server. It is a production dual boot machine that participates in a network with both Windows and Ubuntu machines, some of which are also dual boot. When it participates as a windows machine, it has no problems. When it participates as an Ubuntu machine it is a royal PITA, because regardless of all the lame excuses, Ubundu does not deal well with fakeraid.

ymmv

darkod
November 9th, 2012, 11:23 PM
But aren't you skipping over one very important fact: the fakeraid itself seems misconfigured.

Windows has its own way of doing things, and even often ignores issues, which is not true for linux. So very often it can seem that "things just work" in windows while not completely true and until the problem surfaces.

Ubuntu does work on fakeraid. I can't say if it's perfect, but it does recognize and use it. But the fakeraid shouldn't show two raid devices if you have only one array (with and without 1 at the end of the device name).

As for samba and the automount, it has nothing to do with samba, it doesn't do the auto mount. fstab and the OS do it. Fakeraid gets mounted and used even without samba installed. But not a misconfigured one.

If you have another similar machine (the BX6), check it yourself whether it reports one or two dmraid devices. It should be one. And that machine is working as expected for you, right?

At the end, I will say again. I wouldn't try to share ntfs partitions with samba. Only linux filesystems.

Tranas
November 9th, 2012, 11:57 PM
But aren't you skipping over one very important fact: the fakeraid itself seems misconfigured.

Windows has its own way of doing things, and even often ignores issues, which is not true for linux. So very often it can seem that "things just work" in windows while not completely true and until the problem surfaces.

Ubuntu does work on fakeraid. I can't say if it's perfect, but it does recognize and use it. But the fakeraid shouldn't show two raid devices if you have only one array (with and without 1 at the end of the device name).

As for samba and the automount, it has nothing to do with samba, it doesn't do the auto mount. fstab and the OS do it. Fakeraid gets mounted and used even without samba installed. But not a misconfigured one.

If you have another similar machine (the BX6), check it yourself whether it reports one or two dmraid devices. It should be one. And that machine is working as expected for you, right?

At the end, I will say again. I wouldn't try to share ntfs partitions with samba. Only linux filesystems.

I don't think the fakeraid is misconfigured. Ubuntu never touched it - it is confiugured by Promise for a Windows installation and it works as advertised - for years I might add. Samba simply cannot interpret that configuration, even when the fakeraid array is already mounted manually by 12.04. That inability is no proof of misconfiguration.

The Windows vs Ubuntu issue is a non starter. Either Samba can deal with a Windows machine or it cannot. Aint rocket science. Since Linux has so little of the PC market, you have a ways to go before most people care. The "correctness" of Linux makes little sense if it simply does not work as advertised. If it is so simple to mount manually via 12.04 - why not offer a simple checkbox in configuration - like "mount the checked drive on boot"..... because nobody wants to take the time, apparently. Got a GUI that handles editing automount?? Yes - it's call pysdm - and it does not work. The vast majority of users could give rats *** about command line. The real world uses GUI's.

In this situation, Ubuntu and it's related helper applicatons do not work properly with fakeraid. When Windows handles the networking for the fakeraid, everything is ok and it's not misconfigured. When Samba handles the networking for fakeraid, it is a disaster.

It is obvious that 12.04 handles the fakeraid *manually* just fine, networked or not, so the "misconfigured" fakeraid is a red herring, imho.