BUILD Upgrading to server-based ECC system, please advise

[asw2012]

Hi all, I've been using FreeNAS for 2 years now with this system, FreeNAS-9.1.1-RELEASE-x64 (a752d35):
AMD-BASED BUILD
it's been working great up until a few weeks ago. It seems one of the SATA controllers is on the brink/going out. Random removal reported from FreeNAS (I have 8 drives total), either ada4 - ada5 at one time are not mounting or else ata6 - ata7 are doing the same thing (ata4,5,6,7 are on same "controller"). I've shut off this system so I don't lose any important data incase the thing decides to blow up completely.

I've been told/warned that this system is very discouraged because of the non-ecc ram and since the above mentioned is happening, I am going to replace the current hardware with the below:

Supermicro X10SL7-F-O
Intel Xeon E3-1220V3
Hynix HMT31GR7CFR4C-PBD8 (32GB)

Some questions are:
*will this system be pretty much plug-n-play? Will FreeNAS recognize hardware, and find the ZFS (RAIDz2), and all settings and information will be intact?
*will I be able to plug all 8 disks into the one RAID motherboard controller (8x SAS2 (6Gbps) ports via LSI 2308)? Or will I have to use the 4 x SATA 3.0Gb/s + 2 x SATA 6.0Gb/s ports on the motherboard and buy an add-on card: INTEL Expander card ?

Please let me know if I am going down the right path with the hardware, and figuratively speaking, it will be as simple to swap hardware for FreeNAS to recognize and configure accordingly w/o losing my ZFS pool?

I am using this NAS box for streaming videos to two WD Live boxes (sometimes both at same time) - also storing various media, pictures, files, etc - I am also running a minecraft server.

thank you.
-Andrew

 

[Apollo]

Hi asw2012,

I would say the most challenging part of using the X10SL7-F is dealing with the IPMI interface.
You will need to flash the LSI interface to IT mode version 16 I believe. This is best done through the USB flashing option.
There are resources you can find on this forum as well as others.

Past these steps, you shouldn't have any particular issues. You just need to run your new system with the the same USB key that is running on your AMD, so not to mess with the configs.
You will or may have to reassign your network interface.

I would recommend you run SMART on all the drives and see if you have any Reallocated sector counts and other errors. I suspect your drives may be failing rather than the SATA controllers.
I would try to get a spare drive just in case that could be used to replace a failing drive.

I would do a "scrub" and see if any errors appear (I hope you have your data backed-up somewhere else).
The trouble of using a non ECC system is that it is possible your data or parity checksum may have been corrupted at some point and ZFS would try to fix it by running resilvering of the affected block, hopefully, it is not going to destroy the affected item but it could.
If the drives are too worn out, "scrub" could potentially overstress your pool and cause one or more drives to fail to a point your pool becomes unavailable. Hopefully a reboot should let you reattach the drives, but it would be an indication your pool needs replacing.

I would only upgrade to Freenas 9.3 once the transition to the new hardware as been completed and tested, and due to the nature of the install, I would run Freenas install only with the drives disconnected as not to mistakely write to one of the drive.

Other than that your pool should be accessible with the Supermicro X10SL7-F.

Back to your hardware selection, it seems the RAM is Registered as opposed to Unbuffered so I don't know if the memory will be compatible.

 

[Ericloewe]

Should be painless. Don't upgrade the OS and hardware at once, flash the SAS controller to P16 IT mode and everything should just work.

By the way: SAS expanders only work with SAS controllers, not Intel SATA. Fortunately, you do have SAS ports. Even more fortunately, you don't need an expander.

 

[Z300M]

Hi asw2012,
Back to your hardware selection, it seems the RAM is Registered as opposed to Unbuffered so I don't know if the memory will be compatible.

What he said: if you go to the Supermicro.com Web site, click on that motherboard, then on "recommended memory" (or whatever the wording is), you'll find the part numbers for compatible memory (unbuffered) by Samsung and by Hynix -- and perhaps also Crucial. The Hynix module # is shown in my sig. and works fine.

 

[Ericloewe]

What he said: if you go to the Supermicro.com Web site, click on that motherboard, then on "recommended memory" (or whatever the wording is), you'll find the part numbers for compatible memory (unbuffered) by Samsung and by Hynix -- and perhaps also Crucial. The Hynix module # is shown in my sig. and works fine.

Oh yeah, totally missed that.

OP, DO NOT buy Registered RAM.

 

[Jailer]

The ram you have listed above is not compatible with that motherboard. The Ram you want is this. Notice that it's 16GB and not the 8 you have listed. To do everything you have listed above you'll need more than 8GB of RAM.

Once you have the on board controller flashed to IT mode you can connect all your drives to the ports for the on board controller or a combination of both, Freenas doesn't really care what ports they are on.

 

[asw2012]

Thanks for the feedback. I missed that part with the registered RAM when researching. I won't get registered and am going to use what supermicro recommends, via their link. The flashing is straightforward to P16 IT mode also. I'll upgrade the hardware first, then after all is ok, upgrade the software to FreeNAS 9.3

 

[asw2012]

Finally did the hardware upgrade, used all same hardware as listed above (Supermicro X10SL7-F-O & Intel Xeon E3-1220V3) except used memory: Crucial model CT2KIT102472BD160B (32GB) - Flashing to IT mode was easy.

I'm at FreeNAS 9.1.1
can I jump right into the latest 9.3.1 ?

thanks.

 

[BigDave]

I'm at FreeNAS 9.1.1
can I jump right into the latest 9.3.1 ?

No. I would do a fresh install of 9.3 on a new USB flash or whatever your boot device is,
and import your volume/pool. Don't forget to save a copy of your configuration file beforehand.

 

[Robert Trevellyan]

Consider going via the last update on 9.1 and 9.2.

 

[gpsguy]

Ditto on what @Robert Trevellyan said. If/when you upgrade to 9.3.1, you'll need to flash the controller to P20. Leave it on P16 until then.

 

[BigDave]

Flashing to IT mode was easy.

I just figured the OP would not have an issue after reading that statement. *shrugs shoulders*

 

[gpsguy]

Me too. I just wanted to give him the heads up, that it would be needed. A lot of users were caught off guard, after they upgraded to 9.3.1.

I just figured the OP would not have an issue after reading that statement.

 

[asw2012]

big thanks for the help all
:)

 

[asw2012]

Hope someone can help here. About a week ago had two drives with write errors. I hit the replace disk button on one of them, turned off the box, replaced the cable. I tried cable first too see if it's the cable, not a bad hard drive, I've had this problem before....

I then get the box to boot up again, hit the replace button, pick the drive, then I get this error (attached)

I don't understand what it's trying to tell me...

 

[Robert Trevellyan]

Are you carefully following the disk replacement directions?

I don't understand what it's trying to tell me...

It's trying to tell you that the disk you're using as a replacement is part of an existing pool. Slow down, think carefully and figure out exactly which disk is which. When you're 100% sure that the disk you're using as a replacement is the right one, if you still get the error, you need to wipe the disk and try again.

EDIT: wait a minute, you have or had a pool called RAIDZ2? Good grief.

 

[asw2012]

Are you carefully following the disk replacement directions?

It's trying to tell you that the disk you're using as a replacement is part of an existing pool. Slow down, think carefully and figure out exactly which disk is which. When you're 100% sure that the disk you're using as a replacement is the right one, if you still get the error, you need to wipe the disk and try again.

EDIT: wait a minute, you have or had a pool called RAIDZ2? Good grief.

I've got all the disks sorted so I know which one is which. I did not hit the "offline" before replacing. I think that is the issue. I gotta get to work, I'll futz with this when I'm back home. Thanks for the info.

 

[asw2012]

all is a-ok. thanks for the help