Can't setup RAID5: "Disk has 4KB logical sector size, not yet supported"

355 views
Skip to first unread message

Tomas Black

unread,
Jun 9, 2018, 5:51:54 PM6/9/18
to alt-f
Hello, 


I recently bought a Hitachi 2TB HD, to add on the USB port and mount a RAID5 setup in Alt-F on my DNS320L Rev. A

However, when I go to the Wizard and select RAID5, I get:

"Disk has 4KB logical sector size, not yet supported."


And I'm unable to select it in order to re-format the HDs and proceed....

Please help, what do I do to get RAID5?

Regards, 


Tomas

Tomas Black

unread,
Jun 9, 2018, 6:04:17 PM6/9/18
to alt-f
Additional info: just upgraded it to v. 1.0, from RC 0.4

Upgrade went fine, everything running smoothly. But can't setup RAID5... 

João Cardoso

unread,
Jun 9, 2018, 7:11:51 PM6/9/18
to al...@googlegroups.com


On Saturday, 9 June 2018 22:51:54 UTC+1, Tom wrote:
Hello, 


I recently bought a Hitachi 2TB HD, to add on the USB port and mount a RAID5 setup in Alt-F on my DNS320L Rev. A

However, when I go to the Wizard and select RAID5, I get:

"Disk has 4KB logical sector size, not yet supported."

Yes, that is a Alt-F limitation, as I don't have one such disk to test with.
There was a 3TB disk donation to solve that limitation (thanks Rolf) but it turns out that the disk had normal 512 bytes logical sectors.

4K logical sector disks aren't/wasn't usual on the desktop marked, they were only available for the enterprise/professional market, and as such were expensive.

That said, from your words it looks like you attached the disk to a USB adapter, and some USB adapters sometimes "lie" about the disk logical sector size (search the forum for that, I remember addressing that question EDIT: found it: https://groups.google.com/d/msg/alt-f/FKc-19I8vgo/XMg8PrssMusJ
Have you tried to attach the disk directly to one of the NAS bays? (or is it a 2.5" disk?) Or use another USB enclosure or adapter? I'm afraid that it is your only option. What is the disk model?

As for RAID5, be certain that you "stage" the setup for a few days (including power cuts!) before relying on it, as USB is prone to connect/disconnect/timeout issues (use short and good quality USB cables) and that will require lengthy RAID rebuilds.

Tomas Black

unread,
Jun 9, 2018, 9:32:32 PM6/9/18
to alt-f


Em sáb, 9 de jun de 2018 20:11, João Cardoso <whoami...@gmail.com> escreveu:


That said, from your words it looks like you attached the disk to a USB adapter, and some USB adapters sometimes "lie" about the disk logical sector size (search the forum for that, I remember addressing that question).
Have you tried to attach the disk directly to one of the NAS bays? (or is it a 2.5" disk?) Or use another USB enclosure or adapter? I'm afraid that it is your only option. What is the disk model?

As for RAID5, be certain that you "stage" the setup for a few days (including power cuts!) before relying on it, as USB is prone to connect/disconnect/timeout issues (use short and good quality USB cables) and that will require lengthy RAID rebuilds.


It is a Hitachi 2tb Ultrastar, model HUA7230ALA640, the cheapest 2tb I could buy new. I don't know if it has 4kb logical sectors, need to research its specs.

As for the USB enclosure, it is a Seagate USB3.0 one. Good quality, short USB cable. Previously I had another Hitach on it, HUA7200ALA330, that worked w/o problems.

Tomas Black

unread,
Jun 9, 2018, 9:44:54 PM6/9/18
to alt-f

Tomas Black

unread,
Jun 9, 2018, 9:48:49 PM6/9/18
to alt-f
And this is the USB enclosure 


images (1).jpeg

João Cardoso

unread,
Jun 10, 2018, 11:05:17 AM6/10/18
to al...@googlegroups.com
As said, Alt-F don't manage 4K logical sectors disks from the Disk Wizard or Disk Partitioner. If the "offending" disk is partitioned using the command line, it will be/should be handled fine.

So you have to identify the disk first. I believe that it is the USB attached one. Use the Disk Partitioner and select one disk at a time, the message "Disk has 4KB logical sector size, use the command line to partition it." will be displayed for the relevant disk. Notice its device name, it should be 'sdc'.

If it is the USB disk, unplug it and create a degraded RAID5 on the Disk Wizard using the other two disks.
After that, plug the USB disk and notice its device name, it should be 'sdc', and the commands bellow *assume* that.
You can then use the following commands on the command line after ssh or telnet the box and login as the 'root' user using the same password as in the webUI:

sgdisk --zap-all /dev/sdc # clear disk partition table
sgdisk --largest-new=1 /dev/sdc # create a partition covering the whole disk
sgdisk --typecode=1:fd00 /dev/sdc # change partition type to RAID
sgdisk --print /dev/sdc # prints the disk partition table, for confirmation
sfdisk --re-read /dev/sdc # try to force the kernel to re-read the partition table without a reboot

Please save and post all the above commands outputs for diagnose if any issue arises

Then add it to the already created RAID5 using Disk->RAID, and under "Component Operations", selecting "sdc1" under "Partition" and "add" under "Operation"

After the rebuild finishes, which can take hours, power off the box and power it on again to see if everything is OK.
The USB attached disk must always be powered on and ready before the NAS, to preserve the device names (sdc) and don't let the RAID to be created in degraded mode, which would imply a lengly rebuild.

Worked?


On Sunday, 10 June 2018 02:48:49 UTC+1, Tom wrote:
And this is the USB enclosure 



Em sáb, 9 de jun de 2018 22:44, Tomas Black escreveu:
Em sáb, 9 de jun de 2018 22:32, Tomas Black  escreveu:

Tomas Black

unread,
Jun 10, 2018, 3:46:24 PM6/10/18
to al...@googlegroups.com
Hello João, 


I believe it worked. See the output of the commands:

[root@NAS]# sgdisk --zap-all /dev/sdc
Creating new GPT entries.
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.

[root@NAS]# sgdisk --largest-new=1

[root@NAS]# sgdisk --print /dev/sdc
Creating new GPT entries.
Disk /dev/sdc: 488378645 sectors, 1.8 TiB
Logical sector size: 4096 bytes
Disk identifier (GUID): F5186145-A4DF-4BD6-97E5-8236BD9A5BA1
Partition table holds up to 128 entries
First usable sector is 6, last usable sector is 488378639
Partitions will be aligned on 256-sector boundaries
Total free space is 488378634 sectors (1.8 TiB)

Number  Start (sector)    End (sector)  Size       Code  Name



I did everything else you said, but sdc is not yet showing as availble drive to be added to the RAID5 setup. 

See all SSs I took during the process:



On Sun, Jun 10, 2018 at 12:05 PM, João Cardoso <whoami...@gmail.com> wrote:
As said, Alt-F don't manage 4K logical sectors disks from the Disk Wizard or Disk Partitioner. If the "offending" disk is partitioned using the command line, it will be/should be handled fine.

So you have to identify the disk first. I believe that it is the USB attached one. Use the Disk Partitioner and select one disk at a time, the message "Disk has 4KB logical sector size, use the command line to partition it." will be displayed for the relevant disk. Notice its device name, it should be 'sdc'.

If it is the USB disk, unplug it and create a degraded RAID5 on the Disk Wizard using the other two disks.
After that, plug the USB disk and notice its device name, it should be 'sdc', and the commands bellow *assume* that.
You can then use the following commands on the command line after ssh or telnet the box and login as the 'root' user using the same password as in the webUI:

sgdisk --zap-all /dev/sdc # clear disk partition table
sgdisk --largest-new=1 /dev/sdc # create a partition covering the whole disk
sgdisk --print /dev/sdc # prints the disk partition table, for confirmation

Please save and post all the above commands outputs for diagnose if any issue arises

Then add it to the already created RAID5 using Disk->RAID, and under "Component Operations", selecting "sdc1" under "Partition" and "add" under "Operation"

After the rebuild finishes, which can take hours, power off the box and power it on again to see if everything is OK.
The USB attached disk must always be powered on and ready before the NAS, to preserve the device names (sdc) and don't let the RAID to be created in degraded mode, which would imply a lengly rebuild.

Worked?

On Sunday, 10 June 2018 02:48:49 UTC+1, Tom wrote:
And this is the USB enclosure 



Em sáb, 9 de jun de 2018 22:44, Tomas Black escreveu:

Em sáb, 9 de jun de 2018 22:32, Tomas Black  escreveu:


Em sáb, 9 de jun de 2018 20:11, João Cardoso escreveu:


That said, from your words it looks like you attached the disk to a USB adapter, and some USB adapters sometimes "lie" about the disk logical sector size (search the forum for that, I remember addressing that question).
Have you tried to attach the disk directly to one of the NAS bays? (or is it a 2.5" disk?) Or use another USB enclosure or adapter? I'm afraid that it is your only option. What is the disk model?

As for RAID5, be certain that you "stage" the setup for a few days (including power cuts!) before relying on it, as USB is prone to connect/disconnect/timeout issues (use short and good quality USB cables) and that will require lengthy RAID rebuilds.


It is a Hitachi 2tb Ultrastar, model HUA7230ALA640, the cheapest 2tb I could buy new. I don't know if it has 4kb logical sectors, need to research its specs.

As for the USB enclosure, it is a Seagate USB3.0 one. Good quality, short USB cable. Previously I had another Hitach on it, HUA7200ALA330, that worked w/o problems.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.
Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

João Cardoso

unread,
Jun 10, 2018, 4:27:04 PM6/10/18
to Alt-F


On Sunday, 10 June 2018 20:46:24 UTC+1, Tom wrote:
Hello João, 


I believe it worked. See the output of the commands:

[root@NAS]# sgdisk --zap-all /dev/sdc
Creating new GPT entries.
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.

[root@NAS]# sgdisk --largest-new=1

Missing the device /dev/sdc

 

[root@NAS]# sgdisk --print /dev/sdc
Creating new GPT entries.
Disk /dev/sdc: 488378645 sectors, 1.8 TiB
Logical sector size: 4096 bytes
Disk identifier (GUID): F5186145-A4DF-4BD6-97E5-8236BD9A5BA1
Partition table holds up to 128 entries
First usable sector is 6, last usable sector is 488378639
Partitions will be aligned on 256-sector boundaries

ah... my mistake, another sgdisk command to set the alignment is missing, I will add it shortly.
 
Total free space is 488378634 sectors (1.8 TiB)

Number  Start (sector)    End (sector)  Size       Code  Name

Here the partition "1" should appear 
 



I did everything else you said, but sdc is not yet showing as availble drive to be added to the RAID5 setup. 

I have meanwhile added another command to set the partition type to RAID, then it will appear as available
 

See all SSs I took during the process:

There are some inconsistencies, in some pics raid1 appears, in another raid5...

Please redo all the sgdisk commands on sdc. You don't need to recreate the RAID5 (if it is indeed raid5)

Tomas Black

unread,
Jun 10, 2018, 6:38:53 PM6/10/18
to alt-f
ok, I'll redo all the commands, but what are the new command (missing) that you will add shortly?

I just copy and pasted your commands. (yes, it is indeed /dev/sdc, the USB HD)

Please provide the updated commands, I`ll run them again

Thanks!

Joao Cardoso

unread,
Jun 10, 2018, 6:50:37 PM6/10/18
to Alt-F Group


On Sun, Jun 10, 2018, 23:38 Tomas Black wrote:
ok, I'll redo all the commands, but what are the new command (missing) that you will add shortly?

I have edited the forum post with the new commands.

On second thoughts, don't use the one that sets the alignment. I can't edit the post now, I will do it tomorrow.


I just copy and pasted your commands.

I think that you missed the device name in the command that creates the partition

(yes, it is indeed /dev/sdc, the USB HD)

Please provide the updated commands, I`ll run them again

Thanks!
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Tomas Black

unread,
Jun 10, 2018, 10:23:39 PM6/10/18
to alt-f
Thanks, here is the updated output of the commands:


[root@NAS]# sgdisk --zap-all /dev/sdc
Creating new GPT entries.
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.

[root@NAS]# sgdisk --set-alignment=8 /dev/sdc
Creating new GPT entries.

[root@NAS]# sgdisk --largest-new=1 /dev/sdc
Creating new GPT entries.
The operation has completed successfully.

[root@NAS]# sgdisk --typecode=1:fd00 /dev/sdc
The operation has completed successfully.

[root@NAS]# sgdisk --print /dev/sdc
Disk /dev/sdc: 488378645 sectors, 1.8 TiB
Logical sector size: 4096 bytes
Disk identifier (GUID): 61D7D557-134B-4F50-B441-058779127411
Partition table holds up to 128 entries
First usable sector is 6, last usable sector is 488378639
Partitions will be aligned on 256-sector boundaries
Total free space is 250 sectors (1000.0 KiB)

Number  Start (sector)    End (sector)  Size       Code  Name
   1             256       488378639   1.8 TiB     FD00

[root@NAS]#

I think I'm ready to add it to the RAID5 setup, as soon as the recover process finishes (30min to go)

To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Tomas Black

unread,
Jun 10, 2018, 11:00:27 PM6/10/18
to alt-f
It failed to add:



To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.
20180610_235849.jpg

Tomas Black

unread,
Jun 11, 2018, 11:49:44 AM6/11/18
to alt-f
I rebooted the box and then sdc was added correctly :-)

However, Alt-F shows it with 250Gb only, instead of the 1.8Gb like the others:







My understanding is that it should have 3.6Gb free. Alt-F shows it with only 1.8Gb:





The RAID section of Alt-F shows all drives included on the pool:






What am I missing? Where is the error?

It failed to add:



To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+unsubscribe@googlegroups.com.

João Cardoso

unread,
Jun 11, 2018, 2:03:44 PM6/11/18
to Alt-F


On Monday, 11 June 2018 16:49:44 UTC+1, Tom wrote:
I rebooted the box and then sdc was added correctly :-)

Yes, reboot is one of the possible hypothesis (not loved by linux users :-)
 

However, Alt-F shows it with 250Gb only, instead of the 1.8Gb like the others:

The page also says in red that 4K logical sectors aren't supported.
The displayed partition value was calculated using 512 bytes sectors, if you multiply it by 8 you will get 1.8TB.
The definitive answer is the output of 'sgdisk -p /dev/sdc' that displayed 1.8TB:

[root@NAS]# sgdisk --print /dev/sdc
Disk /dev/sdc: 488378645 sectors, 1.8 TiB
Logical sector size: 4096 bytes
...
Total free space is 250 sectors (1000.0 KiB)

Number  Start (sector)    End (sector)  Size       Code  Name
   1             256       488378639   1.8 TiB     FD00





My understanding is that it should have 3.6Gb free.

Yes, for RAID5: 3 * 1.8TiB / 2 = 3.6TiB

Alt-F shows it with only 1.8Gb:





The RAID section of Alt-F shows all drives included on the pool:



Yes, however, it is displayed in green, which means it is a spare disk, not being used by the RAID. The RAID5 should be displayed as degraded, and a resync should automatically happens, and at that point the green becomes black. A failed component appears in red. A missed componente appears in... well, it doesn't appears ;-)
 



What am I missing? Where is the error?

Can't tell, only using the command line could the situation be diagnosed. "Details" and "Examine" in the RAID Maintenance page (under RAID and Components Operations) would give such information. You can use get it post it.
But it is better to start fresh. I advise you to restart creating the RAID5 again using the RAID Maintenance webUI. As the USB disk is now partitioned you don't have to reuse the command line for that.

Start unmounting the md0 filesystem in the Filesystem Maintenance page (Operations); then in the RAID Maintenance, destroy the md0 RAID (RAID Operations); then in the RAID Creation top section specify the raid type (raid5), each of its three components (sda2, sdb2, sdc1) and Create a new RAID. It will take several hours to rebuild.

If that fails, I can only attribute it to the SATA-USB adapter lying to the OS saying that the enclosure disk has 4K bytes sectors, which certainly is not true for a 2TB disk, and I don't have a solution for it. Please see this post  and the second answer to this one.
 
The Disk Wizard says about RAID5: "Data security and space, with two disks plus an external USB disk (raid5). Complex maintenance."

let us know the outcome


It failed to add:



Tomas Black

unread,
Jun 11, 2018, 4:42:23 PM6/11/18
to alt-f
Ok, I did unmount and destroy the current RAID setup. Then I formatted sdc with the same FS than sda and sdb (btffs). Then I created a new RAID5 setup using the 3 devices. 

It seems it went fine, as now it reports 3.76Gb available, but its status is DEGRADED:





and






However I don`t see any rebuild in progress...... nor on the STATUS option or on the RAID Maintenance option.

See the option DETAILS for md0: 

/dev/md0:
        Version : 1.0
  Creation Time : Mon Jun 11 20:31:55 2018
     Raid Level : raid5
     Array Size : 3905974272 (3725.03 GiB 3999.72 GB)
  Used Dev Size : 1952987136 (1862.51 GiB 1999.86 GB)
   Raid Devices : 3
  Total Devices : 3
    Persistence : Superblock is persistent

  Intent Bitmap : Internal

    Update Time : Mon Jun 11 20:32:46 2018
          State : clean, FAILED
 Active Devices : 1
Working Devices : 2
 Failed Devices : 1
  Spare Devices : 1

         Layout : left-symmetric
     Chunk Size : 512K

           Name : NAS:0  (local to host NAS)
           UUID : faf83c95:090ff181:e202e553:9ce381f1
         Events : 11

    Number   Major   Minor   RaidDevice State
       0       8        2        0      active sync   /dev/sda2
       1       0        0        1      removed
       2       0        0        2      removed

       1       8       18        -      faulty spare   /dev/sdb2
       3       8       33        -      spare   /dev/sdc1

/proc/mdstat:
Personalities : [linear] [raid0] [raid1] [raid6] [raid5] [raid4] 
md0 : active raid5 sdc1[3](S) sdb2[1](F) sda2[0]
      3905974272 blocks super 1.0 level 5, 512k chunk, algorithm 2 [3/1] [U__]
      bitmap: 0/15 pages [0KB], 65536KB chunk

unused devices:


this is the "Examine" option for sdc:

/dev/sdc1:
          Magic : a92b4efc
        Version : 1.0
    Feature Map : 0x1
     Array UUID : faf83c95:090ff181:e202e553:9ce381f1
           Name : NAS:0  (local to host NAS)
  Creation Time : Mon Jun 11 20:31:55 2018
     Raid Level : raid5
   Raid Devices : 3

 Avail Dev Size : 3907026800 (1863.02 GiB 2000.40 GB)
     Array Size : 7811948544 (3725.03 GiB 3999.72 GB)
  Used Dev Size : 3905974272 (1862.51 GiB 1999.86 GB)
   Super Offset : 3907027056 sectors
          State : clean
    Device UUID : eb577339:f1f108ec:874f2aa4:98af21f9

Internal Bitmap : -8 sectors from superblock
    Update Time : Mon Jun 11 20:32:46 2018
       Checksum : 72eaf55d - correct
         Events : 11

         Layout : left-symmetric
     Chunk Size : 512K

   Device Role : spare
   Array State : A.. ('A' == active, '.' == missing)



looks identical to sda and sdb


Disk Utillities Menu:




João Cardoso

unread,
Jun 11, 2018, 8:55:26 PM6/11/18
to Alt-F


On Monday, 11 June 2018 21:42:23 UTC+1, Tom wrote:
Ok, I did unmount and destroy the current RAID setup. Then I formatted sdc with the same FS than sda and sdb (btffs).

Noo, noo, noo ;-)
A RAID is a *device*, like a disk. A filesystem is created on top of a device, not the contrary.
The order is: build the raid, create a filesystem on it, put files on the filesystem.

Creating a RAID on disks or partitions (which are also devices) that already have a filesystem on them (and so possibly contains data) most probably will ruin the filesystem (and the data that is holds).
Be sure that no filesystem is mounted and in use (Disk->Filesystems)  before creating the RAID (although the RAID creation process tries to unmount any component filesystem that is mounted).

But that doesn't yet matter, as your raid device is not OK. When it will be OK you will have to create a filesystem on it. Then you can mount and put data on it.

Then I created a new RAID5 setup using the 3 devices. 

It seems it went fine,

Not really. Somehow sdb2 failed during the assembly, as you can see in the md0 Detail: "faulty spare". And sdc1 is yet a spare.
That is the reason why in the RAID webpage sdb2 appears with the red color and sdc1 in the green color.
Have you read the online help about RAID and filesystems? Hit the blue (?) icon and you will get a mini tutorial and help about the subject and webpage.

Why did sdb2 failed I don't know. It can be in the system logs. And as there are not enough active components, that is probably why sdc1 is still a spare and rebuild didn't start -- can't do it with only a single component.
BTW, your internal disks are very hot, better adjust the fan speed, see Services->System, sysctrl, Configure, and be sure that there is enough free space around the box to exhaust the hot air -- opening the top box cover does *not* help).

Are the internal disks new? I would do a SMART long test on them first Disk->Utilities->Health (after solving the temperature problem).
After that, regarding the RAID, you could try to remove, then clear, then add the faulty component (RAID Operations), but as you don't yet have any data on it (you need to create a filesystem on the raid first), the faster option is to destroy and create the raid again.

PS-You might think that you still have data on the disks, as the used raid format (1.0) is put at the disk end, not destroying filesystems, but that is illusory.
PPS-All your issues shouldn't happen, but it might be good that they happened before you put any data on the RAID5, so you get introduced to the concepts and its intricacies...

Tomas Black

unread,
Jun 12, 2018, 7:31:59 AM6/12/18
to alt-f
Hi João, 


yes I know all of that. I tried to resume the many actions I did to get it working, so it looked dumb on your eyes :-)

I even tried to "level" the FS of sdc to the same of sda and sdb, that's why I said  "I formatted sdc with the same FS than sda and sdb (btffs)". You know, at some point you try starting unusual things to have things working...

In summary: I could not get it going yet. I tried many thing which are too long to describe here, but in short I erased all FSs from all drives, tried partitioning manually them, changed GPT <--> MBR and used quite all optoins of the Atl-F FS Maitennace window :-)


The drives are quite new, not many hours of usage, SMART is OK for all of them, however I have not tried the long test. 

Thanks for the tip regarding the temperature, I oversaw that. Now I've set it to auto and it lowered to high 30s

I have one question for you: how do I setup a swap partition using sgdisk command? The MAN page is not clear on that, at best I could create a swap + RAID partition, but entering the sectores number (which I don't know for this HD), using the    --new=partnum:start:end    option. But it did not look good. I aim to have all the 3 HDs with the exate same space for swap and RAID partitions. 

Another question I have is: for Alt-F, should we use them as GPT or MBR?

Right now I'm starting from scratch. I'm considering partitioning all of them from the command line again (since sdc won't partition from the GUI), identically. Appreciate if you can share some words on this. 

Thanks a lot for your time. By the way, if you want to do other tests with this 4k HD, please let me know. I don't have any data on the HDs :-D


Tom

Tomas Black

unread,
Jun 12, 2018, 7:48:03 AM6/12/18
to alt-f
One additional comment:


you said  "If that fails, I can only attribute it to the SATA-USB adapter lying to the OS saying that the enclosure disk has 4K bytes sectors, which certainly is not true for a 2TB disk"

--> I had another 2TB on this same USB enclosure that was not detected as having 4k bytes sectors. (it was another Hitachi model, that was already failing, I just used it for testing purposes)

So I conclude the problem is a combination  of the HD + USB enclosure. Some work, others don't. Agree?

João Cardoso

unread,
Jun 12, 2018, 2:58:31 PM6/12/18
to Alt-F


On Tuesday, 12 June 2018 12:31:59 UTC+1, Tom wrote:
Hi João, 


yes I know all of that. I tried to resume the many actions I did to get it working, so it looked dumb on your eyes :-)

I even tried to "level" the FS of sdc to the same of sda and sdb, that's why I said  "I formatted sdc with the same FS than sda and sdb (btffs)". You know, at some point you try starting unusual things to have things working...

In summary: I could not get it going yet. I tried many thing which are too long to describe here, but in short I erased all FSs from all drives, tried partitioning manually them, changed GPT <--> MBR and used quite all optoins of the Atl-F FS Maitennace window :-)


The drives are quite new, not many hours of usage, SMART is OK for all of them, however I have not tried the long test. 

Thanks for the tip regarding the temperature, I oversaw that. Now I've set it to auto and it lowered to high 30s

I have one question for you: how do I setup a swap partition using sgdisk command? The MAN page is not clear on that, at best I could create a swap + RAID partition, but entering the sectores number (which I don't know for this HD), using the    --new=partnum:start:end    option.

sgdisk --help is your frient. It has a couple of handy commands, such as --first-in-largest,--first-aligned-in-largest,  --largest-new, etc
 
But it did not look good. I aim to have all the 3 HDs with the exate same space for swap and RAID partitions. 

You don't need, and it isn't recommended, to use swap on USB. The limited USB speed will be shared with the normal disk access, which will make things much slower.

Another question I have is: for Alt-F, should we use them as GPT or MBR?

Doesn't matter. GTP has to be used for greater than 2TB disks. I gave the commands using GPT because they are easier.
As your USB disk is smaller then 2TB you can use the iterative command line 'fdisk' to setup the partitions as you which. Wait... I don't know if 'fdisk' supports 4K logical sectors, probably not.
 

Right now I'm starting from scratch. I'm considering partitioning all of them from the command line again (since sdc won't partition from the GUI), identically. Appreciate if you can share some words on this. 

Thanks a lot for your time. By the way, if you want to do other tests with this 4k HD, please let me know. I don't have any data on the HDs :-D

Thanks
 


Tom

João Cardoso

unread,
Jun 12, 2018, 3:09:47 PM6/12/18
to Alt-F


On Tuesday, 12 June 2018 12:48:03 UTC+1, Tom wrote:
One additional comment:


you said  "If that fails, I can only attribute it to the SATA-USB adapter lying to the OS saying that the enclosure disk has 4K bytes sectors, which certainly is not true for a 2TB disk"

--> I had another 2TB on this same USB enclosure that was not detected as having 4k bytes sectors. (it was another Hitachi model, that was already failing, I just used it for testing purposes)

So I conclude the problem is a combination  of the HD + USB enclosure. Some work, others don't. Agree?

Sometimes manufacturers changes internal circuitry even if given to the user the same brand model. Look at the DNS-323 -- its has three incompatible hardware boards, rev-A1. rev-B1 and rev-C1. Many other consumer devices (e.g., wireless USB adapters) suffer from the same issue.The same might be happening to you.

But the real issue here is the 4K logical block size reported by the USB disk. You should be able to create a degraded RAID5 with the two internal disks. If latter adding the correctly hand-partitioned USB disk fails, than that is the issue -- the 4K logical (reported) disk size and the translation done by the SATA<->USB adapter.

I still think that the inner disk is 512 bytes and the USB adapter is lying. As the links I gave earlier indicate. The only way to be sure about this is to open the USB enclosure, remove the disk and plug it directly to the box bays (or to a generic PC)

 

On Tue, Jun 12, 2018 at 8:31 AM, Tomas Black wrote:
Hi João, 


yes I know all of that. I tried to resume the many actions I did to get it working, so it looked dumb on your eyes :-)

I even tried to "level" the FS of sdc to the same of sda and sdb, that's why I said  "I formatted sdc with the same FS than sda and sdb (btffs)". You know, at some point you try starting unusual things to have things working...

In summary: I could not get it going yet. I tried many thing which are too long to describe here, but in short I erased all FSs from all drives, tried partitioning manually them, changed GPT <--> MBR and used quite all optoins of the Atl-F FS Maitennace window :-)


The drives are quite new, not many hours of usage, SMART is OK for all of them, however I have not tried the long test. 

Thanks for the tip regarding the temperature, I oversaw that. Now I've set it to auto and it lowered to high 30s

I have one question for you: how do I setup a swap partition using sgdisk command? The MAN page is not clear on that, at best I could create a swap + RAID partition, but entering the sectores number (which I don't know for this HD), using the    --new=partnum:start:end    option. But it did not look good. I aim to have all the 3 HDs with the exate same space for swap and RAID partitions. 

Another question I have is: for Alt-F, should we use them as GPT or MBR?

Right now I'm starting from scratch. I'm considering partitioning all of them from the command line again (since sdc won't partition from the GUI), identically. Appreciate if you can share some words on this. 

Thanks a lot for your time. By the way, if you want to do other tests with this 4k HD, please let me know. I don't have any data on the HDs :-D


Tom

Tomas Black

unread,
Jun 12, 2018, 4:09:30 PM6/12/18
to alt-f
Wel... then I can't do much (if the problem is with the USB enclosure). 

Do you know an example of a workng RAID-5? (specially HDs and USB enclosure)

Thanks


Tomas Black

unread,
Jun 13, 2018, 11:15:19 AM6/13/18
to al...@googlegroups.com
Indeed the problem is with the USB enclosure. I swap HDs and here is the screenshot:






the new one (HUA7230020ALA640) is now on the left bay, recognized as sda, and has not the 4kb logical sector error message.

on the other hand, one of the others, now attached via USB, is reported as having 4kb logical sectors...

On, case closed. 


Now the most important question for anyone willing to setup a RAID5 using Alt-F: what external cases work without problems?
image.png

Nicolas Desveaux

unread,
Jun 13, 2018, 2:30:30 PM6/13/18
to al...@googlegroups.com
@Tomas That sounds like such a crazy idea.. I wouldn't dream of setting up any kind of raid over USB...

It failed to add:



To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Visit this group at https://groups.google.com/group/alt-f.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "Alt-F" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alt-f+un...@googlegroups.com.

Tomas Black

unread,
Jun 13, 2018, 3:11:37 PM6/13/18
to alt-f
Hi Nicolas,


Indeed raid5 was the motivator to install alt-f on my DNS-320L :-)

moreno...@gmail.com

unread,
Aug 5, 2018, 4:37:57 AM8/5/18
to Alt-F
Hello,

I had the exact same problem as Tomas and I followed the steps but I coulndn't add /dev/sdc to the degraded RAID5 array.

What I did then is I destroyed the array and created a new one with "Disk -> RAID -> Raid Creation" selecting sda2 sdb2 sdc1 (sd*2 because the internal disks have swap on sd*1) and it took half a day but it worked!

Now though in the main Status window I see under "Disks"

left   sda  HGST HUS724030ALA640 3.0TB    active or idle    37°C/98°F   passed
right sdb  HGST HUS724030ALA640 3.0TB   active or idle    38°C/100°F  passed
usb  sdc  HGST HUS724030ALA640 3.0TB      --                   38°C/100°F  passed

With the last "passed" of usb sdc in red. Why is it in red?

Thanks

Moreno

João Cardoso

unread,
Aug 8, 2018, 11:01:54 AM8/8/18
to Alt-F


On Sunday, 5 August 2018 09:37:57 UTC+1, moreno wrote:
Hello,

I had the exact same problem as Tomas and I followed the steps but I coulndn't add /dev/sdc to the degraded RAID5 array.

What I did then is I destroyed the array and created a new one with "Disk -> RAID -> Raid Creation" selecting sda2 sdb2 sdc1 (sd*2 because the internal disks have swap on sd*1)

Then the sdc existing partition was used.
I really don't know what the long term maintenance or operational consequences are, having a 4096 byes logical sector size disk.
 
and it took half a day but it worked!

Now though in the main Status window I see under "Disks"

left   sda  HGST HUS724030ALA640 3.0TB    active or idle    37°C/98°F   passed
right sdb  HGST HUS724030ALA640 3.0TB   active or idle    38°C/100°F  passed
usb  sdc  HGST HUS724030ALA640 3.0TB      --                   38°C/100°F  passed

With the last "passed" of usb sdc in red. Why is it in red?

From the online help (hit the (?) blue icon near the page title): 

Disks: shows available disks
  • Health: the disk health, as reported by SMART, or unknown, if the drive does not support SMART or is in standby.
  • If you see a message in red, replace the disk as soon as possible, as the disk is likely to fail in the next hours or days.
    If the message appears in blue it means that some disk parameter has failed in the past.
    You can examine the SMART log in Disk->Utilities->Health->Show Status.
Is it a consequence of the 4KB logical sector size? Doubt, but only the full SMART log will help

moreno...@gmail.com

unread,
Aug 17, 2018, 9:31:16 AM8/17/18
to Alt-F
Hello,

the problem is that S.M.A.R.T. on usb does not work properly.

If I query hddtemp from the terminal I get
$ hddtemp
/dev/sda: HGST HUS724030ALA640: 46 C
/dev/sdb: HGST HUS724030ALA640: 45 C
/dev/sdc: Mass Storage Device: S.M.A.R.T. not available

On the Status page I read "--" under "Power Status" of "Disks"

Under Disks->Utilities->Health->Select Action->Show Status I read

=== START OF READ SMART DATA SECTION ===
SMART Status not supported: Incomplete response, ATA output registers missing
SMART overall-health self-assessment test result: PASSED
Warning: This result is based on an Attribute check.

Everthing else seems ok. So nothing really to worry about, I should have imagined that over USB you don't have full control of the disk. It can't spindown or standby and this is a big issue but again, I don't see a way around it.

Hope it helps

Moreno
Reply all
Reply to author
Forward
0 new messages