PLEASE PLEASE HELP!!!

[mcdonsco]

I built a new WHS 2011 Server using an ASUS P8H67-M Pro/CSM mainboard, core i3 2120, 8gb's ram...the works...the RAID 5 is using 4 Seagate 2TB 7,200rpm drives on the built-in ASUS raid controller ("Intel desktop/workstation/server express chipset sata raid controller" is what it comes up as in device manager). Power supply is an Antec eco neo 620w...

Basically this system has been dropping drives out of the array like crazy. Usually random drives but seems to really like dropping the #4 drive most often (and that drive has been replaced for new). It's shown drives out that just needed to be reset to "normal" in intel rapid storage manager, drives that just show up as "missing" status...usually one drive at a time, sometimes two (array would be rebuilding after one went out, then that drive and another one would drop out).

I've been dealing with this for weeks and I'm starting to wonder if it's something with WHS and not the hardware because at this point pretty much ALL of the hardware has been replaced.

Here's what I've tried:
•Reloaded the OS fresh at least 2-3 times
•Re-created the raid array about 3-4 times (once in windows, in rapid storage manager, the rest from the controllers bios)
•Almost all of the drives in the machine have been replaced; maybe one that hasn't been?
•Tested ALL drives using seatools in DOS (bootable seatools disk) AND in windows (smart test, short test, long test and erase test on all four drive - all drives passed, even ones that I've replaced because that particular drive dropped out numerous times)
•Replaced mainboard and power supply
•Replaced all cables to drives (both data and power cables)
•Each time the array is rebuilt

STILL have the same problems.

Is there perhaps a problem with WHS and my hardware? I just can't figure this one out and it's honestly starting to drive me a bit insane.

 

[erpsaa]

Hi Mcdonsco,

I built a new WHS 2011 Server using an ASUS P8H67-M Pro/CSM mainboard, core i3 2120, 8gb's ram...the works...the RAID 5 is using 4 Seagate 2TB 7,200rpm drives on the built-in ASUS raid controller ("Intel desktop/workstation/server express chipset sata raid controller" is what it comes up as in device manager). Power supply is an Antec eco neo 620w...

Basically this system has been dropping drives out of the array like crazy.

Is your board at the link I posted below below? I ask because the Intel 67 express has 2 x SATA3 Ports and 4 x SATA2 ports. The problem is the way ASUS specs it. It claims the chipset, says nothing about any RAID level and if their specs were accurate they would be correct.

These are the storage specs at that link, assuming I did pick your board:

Intel® H67 Express Chipset
2 xSATA 6.0 Gb/s ports
VIA VT6415 PATA controller
1 xUltraDMA 133/100 for up to 2 PATA devices

You cannot build any RAID with those specs, well, maybe RAID-0 with 2 drives, or RAID-1, but not RAID-5 with a DMA PATA controller.

However, ASUS (and I do love em') make lots of Typos, so they placed the actual specs of the 6 storage ports in the back panel I/O section. Not a joke, I've seen typos, but this one is a doozy!

http://usa.asus.com/Motherboards/Intel_Socket_1155/P8H67M_PROCSM/#specifications

Assuming I've not gone crazy (I'm having my own RAID issues today) and that is your board, here is my OPINION:

I will STIPULATE I have a long hostory of bias against what I call Fake RAID, or software/BIOS, mainboard, etc. IMO and others will probably disagree, while others will agree. IMO it's garbage. There is no real controller, it's software buried in the BIOS and the Logical drive (I don't even like calling it that) needs a driver to install the OS. It's not ASUS RAID, it's the IRST. Intel.

Nevertheless, I think I know your issue, but confirm a few things for me since you did an excellent job of troubleshooting

Are these your HDDs?

4 x Seagate 2TB 7200 RPM 64MB Cache SATA 6.0Gb/s 3.5"

(if those are your HDDs) they're dropping off because there are only 2 X SATA3 and 4 x SATA2. Your drives are all SATA3, no? (it does not really matter, I'm simply curious)

However you slice it, you did everything correctly. It's not you, it's the software RAID. It's mixing bandwidth and telling people they can build RAID in this manner.

However, even if the above are not your drives and you have all SATA2, or if you bought 2 x SATA3 AND 2 X SATA2 to try to correct the software RAID bandwidth variance issue, you're still disparate.

With such large storage requirements, it won't pass through. You might get away with it staying at NTFS, but you have ended up with a 3TB GPT RAID-5 array that cannot be handled by mixed bandwidth software RAID.

I can recommend a solution, but first let me know what I messed above and if I've posited any inaccuracies.

I will reserve comment on your Home Server question for now, but if you need that much storage for a server, get ready to dish out lots of $$$$$$ for RACKS of SAS.

Don't go crazy yet, I'm kidding, but let me know where I made mistakes above and where I'm correct so I know exactly what you have and we'll go from there.

Dean

 

[mcdonsco]

that is the board and those are the drives. I have all 4 drives hooked up to the sata 2 controllers (even though they are sata 3, Im not trying to achieve max performance, just fault tolerance / array) and the OS SSD drive hooked up to the sata 3 controller (since I didn't want to mix controller speeds within the array); they are backwards compatible. The raid 5 controller does appear to be an actual intel raid controller, but the asus bios only lets you select ahci, ide or raid for ALL sata ports (including the 2 ata 3 ports).

I'm actually considering just setting up all drives as ahci and doing a software raid in windows to be "done" with this...might try that next?

 

[smorizio]

one thing to check and try..use a smart tool or seatools and check that the firmware of the drives are all the same and up to date. sometime it can be a firmware bug.

 

[erpsaa]

MC,

Gotta jump for now, but:

"(since I didn't want to mix controller speeds within the array); they are backwards compatible. The raid 5 controller does appear to be an actual intel raid controller, but the asus bios only lets you select ahci, ide or raid for ALL sata ports (including the 2 ata 3 ports)."

I know and it's annoying, but I must have missed your mention of the OS drive in one of the SATA3 ports. I have to leave, but let me get this straight so I know when I return:

You of course selected RAID in the UEFI BIOS, however, you installed the Windows OS on a single SSD and then used the software RAID to create a storage RAID-5?

My first assumption was only the 4 x 7,200 HDDs, not a mixed in SSD. Please let me know and I'll reply back and yes, also check the firmware on the SSD as well.

Lastly, what does the SSD read as? IDE?

(Sorry for being confusing, when I say Controller, I mean a Hardware RAID Controller on a PCIe slot).

 

[mcdonsco]

MC,

Gotta jump for now, but:

You of course selected RAID in the UEFI BIOS, however, you installed the Windows OS on a single SSD and then used the software RAID to create a storage RAID-5?

Lastly, what does the SSD read as? IDE?

(Sorry for being confusing, when I say Controller, I mean a Hardware RAID Controller on a PCIe slot).

RAID is selected in BIOS, but the intel controller bios is where I configured the RAID array.

The SSD is just on one of the two sata 3 ports, but ASUS bios only lets you select how ALL 6 sata ports are setup, not one by one, so all 6 (to get RAID capability) are set to RAID...Of course the SSD is not raid so how it's interfacing I have no idea (AHCI/IDE - dunno as again, bios just set to "RAID").

 

[mcdonsco]

one thing to check and try..use a smart tool or seatools and check that the firmware of the drives are all the same and up to date. sometime it can be a firmware bug.

[strike]I've checked on seagates site and I don't see any firmware updates available...that's usually the first thing I do with a new build, ensure all firmware/bios is updated.[/strike]

EDIT: Took some searching on seagates site but I just found it. Updating now.

 

[erpsaa]

RAID is selected in BIOS, but the intel controller bios is where I configured the RAID array.

The SSD is just on one of the two sata 3 ports, but ASUS bios only lets you select how ALL 6 sata ports are setup, not one by one, so all 6 (to get RAID capability) are set to RAID...Of course the SSD is not raid so how it's interfacing I have no idea (AHCI/IDE - dunno as again, bios just set to "RAID").

mcdonsco,

Sorry for the delayed reply. In an odd way we have similar problems. 😉 This should do this trick, you only have 3 ways to fix this. Here is why this is happening, then the 3, or really for you, probably two choices.

My main Workstation, (not server) I built on the ASUS P9 Pro x79 chipset. The only difference betweem the "storage" inputs on the mainboard you use and mine is I have 8 vs. your 6. The first six (empty) are a match to your, all must be RAID. My extra two are USELESS. 2 x SATA3 Marvel for so silly concept called Drive Caching. All it does, IMO, is get some people who do not know better to, in reality, build a RAID-0 Logical drivce, or if it has any use, a RAID-1 mirrow I would not disagree with.

I have two drives plugged into the Mainboard chips, both in the "GREY" SATA2's. 1 BD/DVR R and a second DVD-R. In general server boards come with maybe 4 x SATA2 if that. They do come with pleanty of PCIe slots and for good reason. The storage system on my Main WS consists of 6 x 240GB (almost 1/2 M IOPS) in a RAID-0 Configuration on an Intel Card, which stock provides 8 x SATA3/SAS via 2x. It can be expanded to 16 SAS. That's gives me a Logical drive of about 1.32TB.

By backup drive is a 2TB WD Caviar Black on a second Hardware Controller Card, with 1 x SATA3 Port. It does nightly "DIFFERENTIAL" backup of the Array. Once I tried placing the HDD backup into the SATA3 port, same place you put your SSD. My BIOS was AHCI, yet two things happened. The backup and restore testing was perfect, using replication, snapshots and of course the best method, creating a bootable disk image, so if my RAID-0 tanks.

However, while there array speed was in line with exactly how I built it (see attachment below) the WD SATA3 2TB HDD was giving me readings of about 80 MG/s. It was IDE, in AHCI mode, it happens when you do not use the ASUS or rather Intel IRST with ANY drive or drives, which do not contain the OS. Here are your choice in order of cost:

1. You can try the Windows software RAID, but the chipset may override it, since you have that SSD with the OS and all 6 plugins as RAID. Doesn't hury to try. Just don't make any bets on it.


2. The cheapest fix it for CERTAIN. Buy a Hardware Controller Card (I'm looking for one myself, because the one I bought (I went cheap) may sit on a PCIe slot, but I realized it was not advertized correctly. It's nothing but a card with a driver. It's like the OCZ Revo line. Great drives, a tad slow, but all work with drivers.

3. For about $500 (plus the cost of a backplane if needed to fit yout drives) Get a Hardware RAID Controller. You can even find one cheaper, since if you could go this way
my suggestion is to forget the one SSD Boot. There's just one rub. You have massive storage needs and I kid you not, servers with your storage needs are built RAID-5 and they keep to NTFS Racks. It's unfortunate, but there is a problem being addressed right now with RAID that gets bigger than 2TBs.

Backup issues. Replication, snaps, or the best method for Workstations, imaging to a single drive and making it bootable forces you (right now) to keep it under 2TBs if you wish to be able to backup your array. I use three workstations, my main with RAID-0, because the work I do requires gargantual speed, from the storage to everything.

My other two have RAID-5 and RAID-6. I do back them up, but my main WS with the RAID-0, is backed up either every 12 hours, or at demand with 1-Internal and 2-added externals. My RAID-0 is safer than the RAID-5 and even the RAID-6 with dual parity, because of the timeline of the external backups I do for them.

I suspect if #1 works, you'll stay with that. If not, number 2 will work, you can keep your SSD on the one PCIe controller (getting it the heck off the mainboard storage ports, then all your ports are open for your RAID-5 server array).

Number 2 is a permanent fix. You will stop dropping off drives, as soon as that SSD is removed from the Mainboard SATA3.

Deal: If you find a good one before me, let me know. Look for FIRMWARE, not drivers.



Good luck!