Hello,

 

I'm having issues with my new unraid setup. I'm using 4.3.3. I used the latest Syslinux. My USB Key is booting fine, it goes through most of the boot but then ends up with the following:

 

  mount: special device /dev/disk/by-label/UNRAID does not exist

  /var/tmp/network.cfg: No such file or directory

  /boot/config/network.cfg: No such file or directory

 

I've quadruple checked the label, i've reformatted and tried several things. I can't avoid that error.

 

I found a lot of posts by Joe L. helping users with similar problems. I'm able to mount the USB drive once booted and i can get things working. but from a fresh boot, network will not work.

 

mount -t vfat -w /dev/sda1 /boot

followed by

/etc/rc.d/rc.local

(to start unRAID manually now that the /boot/config hierarchy exists)

 

If i do that, it works fine. but instead of sda1, i have to use sdb. not sure if thats significant. Also the "/dev/disk/by-label/" directory does not exist, but the /dev/disk/by-id/ does

 

I tried adding  append initrd=bzroot rootdelay=10 acpi=off nolapic noapic. No help.

 

I tried enabling my parellel port. Disabled USB 2.0. Nothing so far that i've read in the forums has helped.

 

I attached my syslog

 

Specs:

Kingston Data Traveler 2gig

Intel 975XBX (latest bios)

Celeron-D 2.66ghz

1gig ECC DDR2

ATI Rage3D 8mb PCI

3x Seagate 1TB

UnRAID 4.3.3 Basic

 

It does look a little peculiar.  Is it possible that you did not extract the unRAID installation files to the flash drive, with the option to preserve sub directories, or whatever it may be called with your zip extraction tool?  If extracted correctly, there should be a folder in the root of the flash called config, and in it should be a number of files, such as disk.cfg, ident.cfg, go, etc.

 

Your flash drive is device sdb.  These device symbols are assigned by the Linux kernel as it finds the drives, sda to the first, sdb to the second, etc.  Your WD was the first drive found, so it is sda, then the Kingston flash drive was found and assigned sdb, then the Seagate, assigned sdc.  Joe just used sda as an example.

 

It may be just a coincidence, but there is a bit of confusion in your syslog, related to the logging of the drive identification and setup, and their partition identification.  Because the kernel uses multiple threads to identify hardware, and each one logs its messages to the same syslog, it is not uncommon to have messages intermingled from multiple sources, and even once in awhile messages that stomp on each other, a set of new messages will interrupt or replace part of the logging of another message.  In your syslog, this causes the occurrence of the following, which includes some really weird lines:

Nov 30 00:06:50 Tower kernel: scsi 5:0:0:0: Direct-Access ATA WDC WD2500JD-00H 08.0 PQ: 0 ANSI: 5
Nov 30 00:06:50 Tower kernel: sd 5:0:0:0: [sda] 488397168 512-byte hardware sectors (250059 MB)
Nov 30 00:06:50 Tower kernel: sd 5:0:0:0: [sda]     --- more sda lines, snipped ---
Nov 30 00:06:50 Tower kernel: sda:*1<5>scsi 0:0:0:0: Direct-Access Kingston DataTraveler 2.0 1.00 PQ: 0 ANSI: 2
Nov 30 00:06:50 Tower kernel: sd 0:0:0:0: [sdb] 3905407 512-byte hardware sectors (2000 MB)
Nov 30 00:06:50 Tower kernel: sd 0:0:0:0: [sdb]     --- more sdb lines, snipped ---
Nov 30 00:06:50 Tower kernel: sdb:*2 sda1
Nov 30 00:06:50 Tower kernel: sd 5:0:0:0: [sda] Attached SCSI disk
Nov 30 00:06:50 Tower kernel: scsi 6:0:0:0: Direct-Access ATA ST31000340AS SD15 PQ: 0 ANSI: 5
Nov 30 00:06:50 Tower kernel: sd 6:0:0:0: [sdc] 1953525168 512-byte hardware sectors (1000205 MB)
Nov 30 00:06:50 Tower kernel: sd 6:0:0:0: [sdc]     ---more sdc lines, snipped ---
Nov 30 00:06:50 Tower kernel: sdc: unknown partition table
Nov 30 00:06:50 Tower kernel: sd 0:0:0:0: [sdb] Attached SCSI removable disk
Nov 30 00:06:50 Tower kernel: usb-storage: device scan complete
Nov 30 00:06:50 Tower kernel: *3 unknown partition table
Nov 30 00:06:50 Tower kernel: sd 6:0:0:0: [sdc] Attached SCSI disk
Nov 30 00:07:02 Tower login[1341]: ROOT LOGIN on `tty1'

 

I have inserted *1, *2, and *3, to point out the points of 'conflict'.  At point1, the sda lines are almost complete, when the sdb lines suddenly begin, interrupting an sda line, which should have been sda: sda1.  The sda1 fragment appears later, interrupting the last sdb line at point2, which should probably have been sdb: unknown partition table, but that fragment does not appear until point3.  (The sdc line with unknown partition table is fine, because the Seagate has probably not been formatted.)  But since there is no fragment of message with an sdb1, and there IS an orphaned fragment unknown partition table, then almost certainly the broken sdb line was sdb: unknown partition table.  That indicates there was a problem finding any partitions on sdb, the flash drive.

 

It could be an MBR issue.  Are you sure you used the -ma option with the most recent version of syslinux?

 

I think you are well aware of it, but just in case, the most comprehensive source of help for problems like these is the USB Flash Drive Preparation wiki page, and also check the Best of the Forums, USB Boot Issues section.

My own flash drive ends up as /dev/sdc.  The current kernel detects it after the two SATA drives on the motherboard controller.

RobJ is correct, I just used /dev/sda as an example.  It is not an issue if the flash drive shows up as /dev/sdb.

 

Everything points to a issue with the udev process detecting the flash drive and creating the /dev/disk/by-label entry.

The error message "mount: special device /dev/disk/by-label/UNRAID does not exist" usually means the label was not present when the drive was formatted.

 

In this case, he says he verified it...  But lets see if it is visible under Linux.

 

Try typing the following two commands and post the output to let us see how the flash drive is partitioned, and how it is labeled.

fdisk -l -u /dev/sdb

 

and

vol_id /dev/sdb1

(note the "1" at the end of the second command, defining the first partition on device /dev/sdb)

 

I'll be curious to see what the output of both commands shows.

 

Joe L.

Gentlemen, thanks for the replies!

 

RobJ,

 

I ended up using a batch file created on the forum "makebootfat" because i wasn't having any success booting up originally. the syslinux command he uses in the batch file is:

 

"syslinux\makebootfat -o usb -Y -L UNRAID -b syslinux\ldlinux.bss -F -c syslinux\ldlinux.sys unraid"

 

and it worked (well besides this issue I'm having )

 

//edit: I've confirmed my USB key maintained the proper directory structure of UNRAID. Config Folder with 5 files and 7 files in the root.

 

//edit2: i want to make it clear, i tried the -ma method and i wasn't able to boot. i also tried the suggestions in the link you provided to no avail. it was this makebootfat batch file that finally made things work for me.

 

Joe L.

 

I've attached a new syslog (i removed the WD drive, and added a seagate) as well as the two commands you asked me to run.

 

Thanks again!  Just to be sure, if i were to redo my USB key, nothing would happen to the data on the drives that are in the system, right? it wouldn't ask to reformat them i hope 

 

Gentlemen, thanks for the replies!

 

RobJ,

 

I ended up using a batch file created on the forum "makebootfat" because i wasn't having any success booting up originally. the syslinux command he uses in the batch file is:

 

"syslinux\makebootfat -o usb -Y -L UNRAID -b syslinux\ldlinux.bss -F -c syslinux\ldlinux.sys unraid"

 

and it worked (well besides this issue I'm having )

 

Joe L.

 

I've attached a new syslog (i removed the WD drive, and added a seagate) as well as the two commands you asked me to run.

 

Thanks again!  Just to be sure, if i were to redo my USB key, nothing would happen to the data on the drives that are in the system, right? it wouldn't ask to reformat them i hope 

 

Stop the array by pressing "Stop" on the management interface, then save a copy of the "config" folder and its contents elsewhere, then copy it back after you load the USB drive.  It will have your entire configuration and you should be able to boot without even needing a new parity calc. (assuming you stop the array BEFORE you make the copy of the config folder)

 

unRAID would not reformat your drives, but it would probably go through a full parity calculation.

 

The "fdisk" output of your flash drive is *VERY* interesting.  It really looks messed up.  There should only be one partition, or at least that is all that is needed by unRAID.

 

If the flash drive had U3 software on it, it will need to be removed.  See the Wiki for tips on how to get the software from the manufacturer sites to do it.

 

As an example, here it the equivalent output from my server (my flash drive is /dev/sdc)  One partition, /dev/sdc1, using the entire drive.

fdisk -l -u /dev/sdc

 

Disk /dev/sdc: 1024 MB, 1024966656 bytes

16 heads, 63 sectors/track, 1986 cylinders, total 2001888 sectors

Units = sectors of 1 * 512 = 512 bytes

 

  Device Boot      Start        End      Blocks  Id  System

/dev/sdc1  *          63    2001887    1000912+  e  W95 FAT16 (LBA)

It appears as if you did an fdisk of /dev/sdc, not sdb.   Isn't /dev/sdb your flash drive as reported by RobJ in the prior post?

 

Edit: I looked in the syslog... this time, the flash drive ended up as /dev/sdc.  However, the partition table in it is really messed up, therefore, it is why your volume label is not found.  In fact, there is an error listed in the syslog:

Dec  2 11:54:00 Tower kernel: scsi 0:0:0:0: Direct-Access    Kingston DataTraveler 2.0 1.00 PQ: 0 ANSI: 2

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] 3905407 512-byte hardware sectors (2000 MB)

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Write Protect is off

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Mode Sense: 0b 00 00 08

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Assuming drive cache: write through

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] 3905407 512-byte hardware sectors (2000 MB)

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Write Protect is off

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Mode Sense: 0b 00 00 08

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Assuming drive cache: write through

Dec  2 11:54:00 Tower kernel:  sdc: unknown partition table

Dec  2 11:54:00 Tower kernel: sd 0:0:0:0: [sdc] Attached SCSI removable disk

Gentlemen, thanks for the replies!

 

RobJ,

 

I ended up using a batch file created on the forum "makebootfat" because i wasn't having any success booting up originally. the syslinux command he uses in the batch file is:

 

"syslinux\makebootfat -o usb -Y -L UNRAID -b syslinux\ldlinux.bss -F -c syslinux\ldlinux.sys unraid"

 

and it worked (well besides this issue I'm having )

 

Joe L.

 

I've attached a new syslog (i removed the WD drive, and added a seagate) as well as the two commands you asked me to run.

 

Thanks again!  Just to be sure, if i were to redo my USB key, nothing would happen to the data on the drives that are in the system, right? it wouldn't ask to reformat them i hope 

 

Stop the array by pressing "Stop" on the management interface, then save a copy of the "config" folder and its contents elsewhere, then copy it back after you load the USB drive.  It will have your entire configuration and you should be able to boot without even needing a new parity calc. (assuming you stop the array BEFORE you make the copy of the config folder)

 

unRAID would not reformat your drives, but it would probably go through a full parity calculation.

 

The "fdisk" output of your flash drive is *VERY* interesting.   It really looks messed up.  There should only be one partition, or at least that is all that is needed by unRAID.

 

If the flash drive had U3 software on it, it will need to be removed.  See the Wiki for tips on how to get the software from the manufacturer sites to do it.

 

As an example, here it the equivalent output from my server (my flash drive is /dev/sdc)  One partition, /dev/sdc1, using the entire drive.

fdisk -l -u /dev/sdc

 

Disk /dev/sdc: 1024 MB, 1024966656 bytes

16 heads, 63 sectors/track, 1986 cylinders, total 2001888 sectors

Units = sectors of 1 * 512 = 512 bytes

 

   Device Boot      Start         End      Blocks   Id  System

/dev/sdc1   *          63     2001887     1000912+   e  W95 FAT16 (LBA)

 

ya, my fdisk definitely looks messed up in comparison. I haven't installed a parity drive yet. Just 2 data drives. ill try redoing the USB key again.. removing U3 if its there. I didn't think it was, but seeing the fdisk.. looks like something is there.

 

Thanks again.

It appears as if you did an fdisk of /dev/sdc, not sdb.   Isn't /dev/sdb your flash drive as reported by RobJ in the prior post?

 

It moved to SDC when i put in the 2nd 1tb Seagate drive. (although i did remove the WD 250gig drive) *shrug*

So as long as i stop the array, copy the config folder from the drive. I can reformat the usbkey and redo this?

No problem with the change from sdb to sdc.

 

Stop the array, copy the config folder elsewhere on your PC. (not a data disk, since they will not be there until they are back in the array after you put the config back)

 

Good Luck.

No problem with the change from sdb to sdc.

 

Stop the array, copy the config folder elsewhere on your PC. (not a data disk, since they will not be there until they are back in the array after you put the config back)

 

Good Luck.

 

Great! Thanks again Joe and Rob. I'll post back after I've tried this again.

You want to re-partition AND re-format... not just re-format.

 

Assuming you have nothing to lose since you were going to re-format anyway, ou could boot up, but do NOT mount the flash drive.Then, assuming it is still /dev/sdc (make sure you get this part right, there is NO undo)

 

use fdisk, or sfdisk, od cfdisk to delete the existing partitions, partition it as desired, and mark it as bootable.

If not familiar, easiest would be cfdisk.  It is an interactive, screen oriented partition tool.  Use it to delete the existing partitons.  Remember to "write" the new partition table before you quit.   You really want one bootable FAT or FAT16 partition.

cfdisk /dev/sdc

 

Once it is partitioned, you can try formatting and loading the files.  The MBR may still work, or you may need to re-load it using syslinux -ma

 

Joe L.

And if none of that works, you may have to use the HP formatting tool.  It is the only thing that worked for some.

Yesssssss!

 

i got rid of the partitions using fdisk, created 1 fat partition. formatted, syslinux'd it, copied my old config back and voila it works!

 

You guys are awesome! thank you so very much! 

Yesssssss!

 

i got rid of the partitions using fdisk, created 1 fat partition. formatted, syslinux'd it, copied my old config back and voila it works!

 

You guys are awesome! thank you so very much! 

Those other partitions must have been confusing things all along.

 

Glad you are up and running.

 

Joe L.

Don't forget to remove that "acpi=off nolapic noapic" from the syslinux.cfg append line, it's not helping and may reduce performance.

Rob, good call! thanks. I've removed it. Still works great.

I just had this issue on the 4.53 build - it turns out it must have been a corrupted download or extraction.  I downloaded the 4.53 files again and it worked fine.  

 

Just thought i would post to have a potential fix documented.

 

Mark