DELL BOSS VD RAID1

Hello,
I have in my hands a dell server with a "dell boss vd" card and two M.2 sata ssd in raid1.
The system (FreeBSD12.1) detects it as a /dev/ada0/
Code:
Geom name: ada0
Providers:
1. Name: ada0
   Mediasize: 480036847616 (447G)
   Sectorsize: 512
   Stripesize: 4096
   Stripeoffset: 0
   Mode: r0w0e0
   descr: DELLBOSS VD
   ident: 2427d781f8c60010
   rotationrate: unknown
   fwsectors: 3
   fwheads: 16
I would like to use it as zfs zil, but I have a doubt.
If one of the two ssd breaks, does the system warn me?
Can I check the health of the raid?
Anyone have experience with this device?
Thanks in advance
Franco
 
Seeing as it appears as ada0, it seems to be exposed as just a standard disk. As it’s designed for an OS boot disk, this probably simplifies guest installation as no drivers are needed. I suspect the raid management will require a separate utility/driver.
 
A quick check on-line shows that it looks somewhat similar to (but with fewer features than) the traditional Dell PERC RAID controllers.
BOSS-S1 is a simple RAID solution card designed specifically for booting a server's operating system. The card supports up to two 6 Gbps M.2 SATA drives. The BOSS-S1 adapter card has a x8 connector using PCIe gen 2.0 x2 lanes, available only in the low-profile and full-height form factors. The BOSS-S1 modular card has a dedicated slot in modular servers.

NOTE: : BOSS-S1 card allows you to create only one virtual disk from the available physical disks. Specifying the size of a virtual disk is not supported.

NOTE: : There are no status LEDs on the BOSS-S1 card.

You can configure a single RAID1 device from the BIOS. And you can then boot from that virtual "disk".

You don't have to use RAID1 in the controller. It also operates in JBOD (non-RAID) mode. Remove the RAID configuration in the BIOS to expose two raw "disks". You would then have the option to configure software RAID in FreeBSD if you wanted (ZFS or gmirror).

I got the impression that hot swapping SSDs was a problem. You might want to investigate this, as any SSD failure would mean an outage.

The controller supports the SMART protocol. This has the potential to monitor and warn of problems. You would need to test it with smartd to see what and how it reports.

There's a separate set of CLI utilities available for Windows, Linux and VMware. I was not able to download them to check further (need a service tag or serial number). The traditional PERC CLI utilities are both useful and extensive.

However, the CLI utilities are not available for FreeBSD. But SMART would probably give you enough for basic error detection.

PERC controllers have a battery. This is essential for preserving the cache in the event of power loss. There is no battery on the BOSS-S1, but the on-board cache is write-through.

TRIM is supported in JBOD (non-RAID) mode. Presumably the RAID controller does the TRIMing when RAID1 is configured in the controller..

SLOGs (dedicated ZILs) mostly need really low latency and reliability when power is lost. Given that the cache is write-through, I think it would work as a SLOG provided:
  1. you configure the SSDs in some sort of mirror (hardware or software); and
  2. the SSDs have power loss protection (the data sheet for the model you have should tell you that).
However SLOGs only need tiny capacity. You are going to have a lot of spare space.

Also, SLOGs only help when you have heavy synchronous writing (databases and the like).
 
Ciao,

You don't have to use RAID1 in the controller. It also operates in JBOD (non-RAID) mode. Remove the RAID configuration in the BIOS to expose two raw "disks". You would then have the option to configure software RAID in FreeBSD if you wanted (ZFS or gmirror).

that's what I thought, but the bios configurator doesn't allow it. At least I haven't been able to show the two ssds to freebsd as individual disks. I broke the raid but the freebsd installation it only presents me ada0.

The controller supports the SMART protocol. This has the potential to monitor and warn of problems. You would need to test it with smartd to see what and how it reports.

Unfortunately...

Code:
smartctl 7.0 2018-12-30 r4883 [FreeBSD 12.1-RELEASE-p1 amd64] (local build)
Copyright (C) 2002-18, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Device Model:     DELLBOSS VD
Serial Number:    2427d781f8c60010
Firmware Version: MV.R00-0
User Capacity:    480,036,847,616 bytes [480 GB]
Sector Sizes:     512 bytes logical, 4096 bytes physical
Device is:        Not in smartctl database [for details use: -P showall]
ATA Version is:   ATA8-ACS, ATA/ATAPI-7 T13/1532D revision 4a
Local Time is:    Fri Jan 31 09:06:41 2020 CET
SMART support is: Unavailable - device lacks SMART capability.

Now I have remote access to the machine, but in the next days I will try again with the configuration of the controller for jbod and let you know.

Thanks
 
I recommend you check the firmware revision in the BOSS-S1 card.

If you have a maintenance agreement, request the smartctl database information from Dell.
 
Ciao,
I have the physical server on my hands, the firmware is updated to 2.5.13.3020 (the download version on the site is 2.5.13.3016), but the only way to show FreeBSD disks is by configuring the raid. If the raid is not configured camcontrol devlist does not show the device. Any idea?

T.i.a.
 
I have never actually seen one of these cards. Please verify that you have a "Dell BOSS-S1" card. The Dell documentation says:
  • For the BOSS-S1 card, unconfigured drives are automatically non-RAID drives. To convert RAID drives to non-RAID drives, delete the virtual disk.
  • The BOSS-S1 only supports one configured (RAID) and two unconfigured states. The BOSS-S1 card does not break unconfigured drives into RAID and non-RAID state
Is there a way to log into the iDRAC, navigate to storage setup, and delete the virtual disk?

If not, when you boot the system, look on the iDRAC console scrolling output for an invitation to configure the RAID setup by pressing one of the function keys, at the appropriate point. Enter the setup and delete the virtual disk.
 
I had read the documentation before buying the hardware for my customer. I expected to see the two disks and to be able to configure them in the zfs mirror.
But the situation is this: if from the controller bios I delete virtual disk and set the disks as unconfigured, FreeBSD does not detect the controller and therefore does not see the disks, if from the controller bios I set the disks in raid1 Freebsd detects the controller and see the disk ada0.
 
RedHat Linux is on the "supported" list. I would boot CentOS and see what sfdisk -s has to say about the unconfigured disks.
 
Also, have you tried both methods mentioned above for changing the RAID configuration (iDRAC disk management GUI, and talk to the BIOS via the console at boot time)?
 
In the console boot sequence I see the controller BOSS-S1 with 2 disk unconfigured. The BIOS NVMe setting is not raid. In iDRAC I try to deploy S.O from iDRAC GUI, select other operating system and boot. Now Freebsd see a device iDRAC OEMDRV with only one disk, DA1. I tried to boot with an ubuntu stick without installing and ubuntu sees the two m.2 disks.
 
Interesting. I reset the bios and iDRAC to the default. Now FreeBSD does not see neither controller nor disks, but ubuntu sees disks without mounting the iDRAC OEMDRV device.
 
My recollection regarding the PERC controllers is that some basic configuration things only work with the BIOS interface using the console. I would not be at all surprised to see this carry through to the BOSS-S1 controllers.

From your observations, my concern is that Dell may have tweaked the Linux disk drivers to be aware of quirks in the BOSS-S1 RAID controller when running in JBOD mode -- and that these tweaks may not be available in FreeBSD.
 
Dell systems have always been unpredictable when enumerating drives. On models with an SD card, it might appear first in the list, or might be last.

Take care to examine da1 (capacity will be the clue), just to be sure of what you are looking at.
 
I have given up. I mirrored the DELL BOSS-S1 with another ssd of the same size connected to the PERC H330.
I have to put the server into production and unfortunately I have little time.
Thanks for the support.
 
wait.. before I tried to install FreeBSD 12.1 release, now I have tried with version 12.1 stable r357836 and magically in camcontrol devlist I see this:
Code:
<AHCI SGPIO Enclosure 2.00 0001> at scbus6 target 0 lun 0 (pass0, ses0)
<AHCI SGPIO Enclosure 2.00 0001> at scbus15 target 0 lun 0 (pass1, ses1)
although unfortunately continuing with the installation I do not see the disks that I expect.
At boot time I have loaded mrsas driver and now I have another problem. The PERC330 controller configured in HBA mode sees sata3 disks as sata1.
 
wait.. before I tried to install FreeBSD 12.1 release, now I have tried with version 12.1 stable r357836 and magically in camcontrol devlist I see this:
Code:
<AHCI SGPIO Enclosure 2.00 0001> at scbus6 target 0 lun 0 (pass0, ses0)
<AHCI SGPIO Enclosure 2.00 0001> at scbus15 target 0 lun 0 (pass1, ses1)
although unfortunately continuing with the installation I do not see the disks that I expect.
At boot time I have loaded mrsas driver and now I have another problem. The PERC330 controller configured in HBA mode sees sata3 disks as sata1.

There is firmware and/or FreeBSD issues with the Dell BOSS controllers. Make sure you aren't using one with a too _new_ firmware. You can download older firmware versions from support.dell.com.

Firmware A01 (2.5.13.2008), A02 (2.5.13.2009) & A03 (2.5.13.3011) seems to work. A04 (2.5.13.3016) too but with some errors/warnings at boot up. A05 (2.5.13.3020) and A06 (2.5.13.3022) doesn't work (atleast not with "raw" Non-raid-configured drives). Atleast not for us...

Btw, we've seen disk "freezes" (SSD stops responding and goes away) on lower firmware versions - but I've not sure if it's controller related or just the same crappy Intel S3520 SSD (firmware (with the latest Dell firmware DL43) problems that we've seen on all variants (M.2 (BOSS), SATA and PCIe) of that SSD.

I've bug reported the BOSS problem, but no response from anyone with knowledge about the driver for these controller (as of yet):

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=243401
 
At boot time I have loaded mrsas driver and now I have another problem. The PERC330 controller configured in HBA mode sees sata3 disks as sata1.

Regarding the PERC H330 in HBA mode - you really want to switch to HBA330 controllers - or reflash the H330 to HBA330 firmware. It's much more stable and you'll get more direct access to the disks without the RAID firmware getting in the way... (same hardware, just different firmware - and FreeBSD drivers (mpr instead of mrsas)).


Crossflashing is totally unsupported by Dell of course... But I've done it and it works :). Just don't do it in a Dell R330/T330 server - failed there. Worked with the H330 installed in a Dell OptiPlex 990 instead)
 
Ciao,
I flashed the controller firmware and now the S.O. sees disks as sata3.
I installed the S.O. in a zfs mirror made by ada0 (DELLBOSS 6gb/s) and da9 (ssd 12gb/s).
can I have problems?
 
Back to the Dell BOSS-S1 problem. With FreeBSD 12.2 and earlier (atleast 11.3) it will fail to detect "unconfigured" disks on the BOSS adapter if it runs firmware 5, 6 or 7. It works fine with configured (RAID0 or RAID1) devices.

I did some kernel debugging tonight and found that with a patch I _can_ get the AHCI device driver to detect the unconfigured disks even with the latest Dell BOSS-S1 firmware. See the attached patch if you're brave enough to test it. I added a "debug.ahci_verbose" sysctl that can be set to 1 to get more verbose output.

With firmware v4 the output looks like this (from an older version of the patch without the timestamp output from the status changed lines):

Code:
ahcich14: AHCI reset...
ahcich14: SATA status changed 00000133
ahcich14: SATA connect time=0us status=00000133
ahcich14: AHCI reset: device found
ahcich14: AHCI reset: device ready after 0ms
ahcich15: AHCI reset...
ahcich15: SATA status changed 00000133
ahcich15: SATA connect time=0us status=00000133
ahcich15: AHCI reset: device found
ahcich15: AHCI reset: device ready after 0ms
ahcich16: AHCI reset...
ahcich16: SATA status changed 00000113
ahcich16: SATA connect time=0us status=00000113
ahcich16: AHCI reset: device found
ahcich16: AHCI reset: device ready after 0ms

and no ada0 / ada1 disks are detected.

With firmware v7 it looks like this:

Code:
ahcich14: AHCI reset...
ahcich14: SATA status changed 00000000
ahcich14: SATA status changed 00000001
ahcich14: SATA status changed 00000133
ahcich14: SATA connect timeout time=212300us status=00000133
ahcich14: AHCI reset: device not found
ahcich15: AHCI reset...
ahcich15: SATA status changed 00000000
ahcich15: SATA status changed 00000001
ahcich15: SATA status changed 00000133
ahcich15: SATA connect timeout time=212000us status=00000133
ahcich15: AHCI reset: device not found
ahcich16: AHCI reset...
ahcich16: SATA status changed 00000000
ahcich16: SATA status changed 00000113
ahcich16: SATA connect time=100us status=00000113
ahcich16: AHCI reset: device found
ahcich16: AHCI reset: device ready after 0ms

It still (incorrectly) claims that "device not found", however...

Code:
ada0 at ahcich14 bus 0 scbus16 target 0 lun 0
ada1 at ahcich15 bus 0 scbus17 target 0 lun 0
pass4 at ahcich16 bus 0 scbus18 target 0 lun 0

The patch basically increases the time the driver loops waiting for the disks to be online by a lot. It seems the BOSS card is busy much longer with the newer firmware.

Patch attached.
 

Attachments

  • ahci.c.patch.txt
    5 KB · Views: 99
Was facing this issue as well with a BOSS-S1 I bought on ebay that was running A06.
Sadly A07, even though it has some nice fixes, still has the issue of now showing the drives without putting them in RAID.

Has anything ever come further of the patch?

A03 worked for me. Now to shut up the fans in this Dell R730xd using ipmitool.
 
Back
Top