Cannot create volume on Freenas 9.10 (9.3)

[zdr]

I am having a painfully hard time with this sw. I am unable to create volume with volume manager, and what is particularly disappointing, there is no error at all. Window simply closes and nothing happens. I tried all available raid combinations, but volume manager refuses to cooperate. Disks were part of a raid before, created by LSI raid controller, but I tried wiping them already in Freenas to no avail. All I achieved is that there are now two labels less in glabel status output. I also tried this first on 9.3 then upgraded to 9.10 with no change in behaviour.

[root@nas70 ~]# gpart show
=> 34 15663037 da0 GPT (7.5G)
34 1024 1 bios-boot (512K)
1058 6 - free - (3.0K)
1064 15662000 2 freebsd-zfs (7.5G)
15663064 7 - free - (3.5K)

=> 34 3907029101 ada0 GPT (1.8T)
34 94 - free - (47K)
128 4194304 1 freebsd-swap (2.0G)
4194432 3902834696 2 freebsd-zfs (1.8T)
3907029128 7 - free - (3.5K)

[root@nas70 ~]# glabel status
Name Status Components
gptid/cd03be4a-8aea-11df-9c35-8843e138c92c N/A da0p1
gptid/74853a13-8b00-11df-b99e-8843e138c92c N/A ada0p1
gptid/74962045-8b00-11df-b99e-8843e138c92c N/A ada0p2
[root@nas70 ~]#

Any help would be appreciated. I am running this on UCS C200M1 with four 2GB SATA disks connected to embedded LSI sata controller in AHCI mode. Tried enhanced IDE as well without luck.

 

[Ericloewe]

Wipe the disks thoroughly and try again. If you're not comfortable doing this yourself in the FreeBSD CLI using dd, there are several tools out there that can do that for you, some are even included in bootable ISOs with lots of useful utilities.

 

[zdr]

Thanks, I will do that. However, I still don't understand how any software can be written in such a way to not present a user with simple error popup, however generic, when error occurs? This is not version 0.99, it went as far as almost 10 now. Is there a log I can pull up to see what is actually happening under the hood?

Secondly, why it even cares what is on the disks? LSI for example, or any other disk manager software for that matter, does not have an issue to wipe out anything that is there in a millisecond.

 

[zdr]

So I issued this command on first disk:

[root@nas70 ~]# dd if=/dev/zero of=/dev/ada0

And I have been waiting for prompt to return 6 hours now. No progress indicator whatsoever.

 

[Mirfster]

Yeah, that is going to take a while. Search the forums, I think that others have used "gpart destroy -F" to do this a lot faster. I have not used it yet personally, so keep that in mind. Maybe others will chime in (like @depasseg who I think has used this command in the past).

 

[danb35]

LSI for example, or any other disk manager software for that matter, does not have an issue to wipe out anything that is there in a millisecond.

Because there have been too many complaints of the form of "FreeNAS wiped out my existing data", so the developers put in "seat belts" to prevent it from doing so. More recently (in 9.10), they've added a checkbox to overwrite the disks anyway.

 

[depasseg]

Yep - gpart is sooo much faster.
https://forums.freenas.org/index.php?threads/volume-manager-doesnt-see-my-3tb-hdd.41099/#post-261630

 

[zdr]

Can't even open the shell any more, it just displays loading and then closes - without error, as usual. Nice.

 

[Mirfster]

Let's backup here.. Exactly what "embedded LSI sata controller" model are we talking about here?

Per the Cisco's site:

Built-in RAID0 and 1 support for up to four SATA drives; RAID 0 and 1 support for up to four SAS or SATA drives with optional mezzanine card; and RAID 0, 1, 5, 6, and 10 support for up to four SAS or SATA drives with optional LSI MegaRAID card

Can't even open the shell any more, it just displays loading and then closes - without error, as usual. Nice.

Does FreeNas boot and function normally without those drives attached?

Disks were part of a raid before, created by LSI raid controller

Try to simply connect them back to the Raid Controller and have it wipe the drives and delete any configuration (if possible).

 

[zdr]

I did many wipes with zeros, and now tried gpart destroy. Nothing helps.

[root@nas70 ~]# gpart show
=> 34 15663037 da0 GPT (7.5G)
34 1024 1 bios-boot (512K)
1058 6 - free - (3.0K)
1064 15662000 2 freebsd-zfs (7.5G)
15663064 7 - free - (3.5K)

[root@nas70 ~]#

After trying to create volume I have this:

[root@nas70 ~]# gpart show
=> 34 15663037 da0 GPT (7.5G)
34 1024 1 bios-boot (512K)
1058 6 - free - (3.0K)
1064 15662000 2 freebsd-zfs (7.5G)
15663064 7 - free - (3.5K)

=> 34 3907029101 ada0 GPT (1.8T)
34 94 - free - (47K)
128 4194304 1 freebsd-swap (2.0G)
4194432 3902834696 2 freebsd-zfs (1.8T)
3907029128 7 - free - (3.5K)

[root@nas70 ~]#

This is on Cisco UCS C200M1 server with LSI SATA controller on motherboard. Freenas worked fine on another server of the same type. I will however compare them thoroughly if you think it might be hardware related.

 

[Mirfster]

What is the output of: sas2flash -list (note you can redact the SAS Address if desired)?

 

[zdr]

Update1: I am unable to wipe disks ada1, ada2 and ada3 from gui, but this time I am getting the error back: not permitted. I can only wipe ada0, and there the gpt partition is getting created as per above output.

Error: Failed to wipe ada1: dd: /dev/ada1: Operation not permitted

I am now (after reboot) also getting an error when trying to create volume: gtpError: Unable to GPT format the disk "ada1"

Update2: checked the controller, and it's actually Intel SATA embedded controller. Having said that, sas2flash does not work.

 

[Mirfster]

Hmm.. Might this be the issue since you have them connected directly to the MB SATA Controller (per the same Cisco site I linked earlier):

Up to four internal SAS or SATA disk drives; up to 2 terabytes (TB) total

Also, I am assuming that you meant "four 2TB SATA"...

I am running this on UCS C200M1 with four 2GB SATA

While the MB may not be able to address much in terms of space, I would presume that a proper HBA should.

So, with that said I would recommend:

1. Get those drives hooked up to a real HBA and try it
   • I am thinking your MB is not sufficient for directly connecting and makes me kind of question the entire system a bit...
2. If needed, connect them to another machine and ensure to wipe them properly. Like use Windows DiskPart with the "Clean" command (will be very fast); but understand what you are doing first so you don't wipe the wrong drives. Heck, just search and download a WinPE ISO and use that to boot the Server in order to run DiskPart...

 

[zdr]

Found a solution for wiping disk ada1:

sysctl kern.geom.debugflags=0x10

then the wipe goes through and volume is finally created.

 

[Mirfster]

Interesting. I recall that setting from "[How To] Hard Drive Burn-In Testing". This setting does not survive a reboot, so I am curios as to how things work once you have rebooted?

 

[zdr]

It does not have to survive rebooting, it's only needed to allow disk wipe. Now that I wiped disk and created volume, setting can be set back to default.

 

[Mirfster]

K, glad to see that things worked out.