Conflicting info 9.3.1

[sonnyinaz6]

Greetings,

I have an HP DL380G6, dual quad 3ghz proces, 98 gb ram, 6 1 tb sata (HP drives) in z1 pool.

I don't leave the server running 24/7, typcially it's up about 4-5 days a week. Today I booted up, and when I logged in I got an error message (see pic) Critical drive issue. However the server hardware reports no such issue, and the lights on all the drives are green, and all show activity. I did a zpool status with the following results:
zpool status
pool: DATA
state: ONLINE
scan: scrub repaired 0 in 1h11m with 0 errors on Sun Jul 12 01:11:12 2015
config:

NAME STATE READ WRITE CKSUM
DATA ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
gptid/d837ac54-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0
gptid/d85f446d-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0
gptid/d8888db1-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0
gptid/d8b109bd-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0
gptid/d8d9a5d3-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0
gptid/d9025545-0c24-11e5-9083-0026552ea84a ONLINE 0 0 0

errors: No known data errors

pool: freenas-boot
state: ONLINE
scan: scrub repaired 0 in 0h2m with 0 errors on Sun Aug 16 03:48:05 2015
config:

NAME STATE READ WRITE CKSUM
freenas-boot ONLINE 0 0 0
gptid/b984c658-0bb4-11e5-86d0-0026552ea84a ONLINE 0 0 0

errors: No known data errors

So I'm a bit confused, the critical errors state the drive is in someway bad, but everything else is contradicting this? Any ideas what I need to do?

 

[Darren Myers]

are you using HW raid, or do you have an HBA connected that the backplane of the hdds is connected too?

 

[sonnyinaz6]

The controller unfortunately doesn't support jbod, so I created a raid 0 on all 6 drives "separately". Each drive is it's own raid 0, so in essence, no raid, the system and freenas both see 6 independent drives.

 

[Darren Myers]

yea, thats bad, very bad for FreeNAS....you have ZFS which wants direct access to the drives, but you have it ontop of HW raid. While FreeNAS may "see" each drive as JBOD is truly isnt and the HW controller can do some funny thing, funny bad things sometimes. Are you able to retrieve SMART data from the hdds?

 

[sonnyinaz6]

By hardware yes, but the hardware is all reporting fine... and running diagnostics, I'm not sure how to get smart info through freenas...
However, the drive in raid 0 writes the same as normal when only 1 drive is in use, so there shouldn't be any funny businesses going on, I'm aware that freenas doesn't like hardware raid...

 

[SweetAndLow]

By hardware yes, but the hardware is all reporting fine... and running diagnostics, I'm not sure how to get smart info through freenas...
However, the drive in raid 0 writes the same as normal when only 1 drive is in use, so there shouldn't be any funny businesses going on, I'm aware that freenas doesn't like hardware raid...

There is lots of funny business going on when using raid 0. The raid controller is buffering writes and changing the way it's written to disk. If you can't change the raid controller without things breaking then your disks aren't being written to correctly.

 

[Darren Myers]

you could try and get the SMART data via

Code:

 smartctl -a /dev/adaX

where X is 0-5 of the drives

 

[sonnyinaz6]

=== START OF INFORMATION SECTION ===
Vendor: COMPAQ
Product: RAID 0
Revision: OK
User Capacity: 1,000,171,331,584 bytes [1.00 TB]
Logical block size: 512 bytes
Logical Unit id: 0x600508b1001c941c68aa9574b1c80976
Serial number: 50123456789ABCDE
Device type: disk
Local Time is: Fri Aug 28 11:58:48 2015 PDT
SMART support is: Unavailable - device lacks SMART capability.

nas say it doesn't, but I can assure you ALL hp server drives support smart.....

 

[Darren Myers]

see that smart output is an issue....and while HP servers and the drives support SMART....it doesnt mean that being in RAID0 for each drive passes the SMART data to FreeNAS....im not sure what that critical error is...but i think i would move data off the current setup (RAID0 drives) and get a actual HBA and then use FreeNAS with that so that FreeNAS has access to the drives directly.

 

[SweetAndLow]

=== START OF INFORMATION SECTION ===
Vendor: COMPAQ
Product: RAID 0
Revision: OK
User Capacity: 1,000,171,331,584 bytes [1.00 TB]
Logical block size: 512 bytes
Logical Unit id: 0x600508b1001c941c68aa9574b1c80976
Serial number: 50123456789ABCDE
Device type: disk
Local Time is: Fri Aug 28 11:58:48 2015 PDT
SMART support is: Unavailable - device lacks SMART capability.

nas say it doesn't, but I can assure you ALL hp server drives support smart.....

This is a problem, your controller doesn't support passing smart data through to the OS. Freenas has no way of knowing the health of the disks. You need to backup your data and get a card that works with zfs.

 

[sonnyinaz6]

454273-001 HD HP 1TB 7.2k SATA II 3.0GB 3.5 Pol.
MAIN SPECIFICATIONS :

• PRODUCT DESCRIPTION : HP MIDLINE - HARD DRIVE - 1 TB - SATA-300
• HARD DRIVE - 1 TB - SATA-300
• TYPE : HARD DRIVE - HOT-SWAP
• FORM FACTOR : 3.5 X 1/3H
• DIMENSIONS (WXDXH) : 4.4 IN X 7.2 IN X 1 IN
• CAPACITY : 1 TB
• INTERFACE TYPE : SERIAL ATA-300
• AVERAGE SEEK TIME : 8.5 MS
• SPINDLE SPEED : 7200 RPM
  
  GENERAL SPECIFICATIONS :
• DEVICE TYPE : HARD DRIVE - HOT-SWAP
• FORM FACTOR : 3.5 X 1/3H
• INTERFACE TYPE : SERIAL ATA-300
• FEATURES : NATIVE COMMAND QUEUING (NCQ)
• 
  
• COMPLIANT STANDARDS : S.M.A.R.T.

• SEEK TIME : 8.5 MS (AVERAGE) / 17 MS (MAX)
• TRACK-TO-TRACK SEEK TIME : 1 MS
• INTERFACES : 1 X SERIAL ATA-300
• MIN OPERATING TEMPERATURE : 41 °F
• MAX OPERATING TEMPERATURE : 131 °F

 

[joeschmuck]

The drive may be SMART compliant however the guys here are telling you that the controller card is not allowing the OS to talk to the drive directly in order to get that SMART data.

 

[sonnyinaz6]

Guess I'll have to locate an hp controller that supports jbod then...
Thanks for the help

 

[Darren Myers]

Guess I'll have to locate an hp controller that supports jbod then...
Thanks for the help

or or, pick up a LSI 9211-8i or a M1015 (which has to be crossflashed)

 

[sonnyinaz6]

Wouldn't know how to connect them, the HP has it's own proprietary connectors to the backplane....

 

[danb35]

Are you sure they're proprietary and not just SFF-8087 mini-SAS?

 

[Darren Myers]

Wouldn't know how to connect them, the HP has it's own proprietary connectors to the backplane....

pretty sure on my DL380 G6 the backplane has 2 SFF 8087 connectors (1 per 4 hdds, totaling 8 hdds) , but my DL 380 G6 uses the SFF. Do you have a picture of your server, the SFF looks like the first image, and the LFF is the second.

 

[sonnyinaz6]

My apologies, I didn't mean to bail on you earlier, I work a 12 hr graveyard and was up 4 hours after being home already and just died about that time. Being that tired didn't help me see the connector any better either. You're right, it is an sff 8087, so it's not an issue with the connectors. So I just have to locate that type matching the other end that whatever lsi I can find has.

 

[sonnyinaz6]

Ok, just ordered the LSI 9211-8i. Had to order another riser card though, that's an 8x pcie card, I had the 4x slots on my current card. But, both ordered, gotta hook up my old thecus and copy everything back to that (that'll take at least as long as the shipping time)

Again, thank you all so much for the help. I just want to get this thing reliable...

 

[Nick2253]

That's one hell of a server you've got there for only 6TB. :D.

If you want to get this thing reliable, I'd also suggest that you go to RAIDZ2. RAIDZ1 (single disk parity) is not a good defense against drive failure. Google "RAID5 is dead" to learn more.

Something to think about, as you go forward, is looking for some newer hardware. I would imagine that you're drawing quite a few watts with that setup, and you'd be able to save quite a bit of money on power by going to something a bit newer. You'd probably be able to sell the DL380G6 for what a decent modern setup would cost you, and then you'd be money in the bank every month with the power bill came due.