Hard Drive timeout NVIDIA NFP3600

[TheDataBox]

Build
Tyan S2937 and Tyan S2912
Dual core and Quad core Opteron Socket F
DDR2-5300 ECC REG 16GB total sharing between boards
5 WD RE4 SATA HDD's / sharing between boards
2 PSU one 550 other 500, differnt brands.
Tried different SATA cables
Both MB BIOS are recent

I went with the MB because it has plenty of sata slots from the chipset "NVIDIA NFP3600" 6 slots at SATA II
The only notable difference between these boards is S2912 will support the new revision Opterons "Shanghai / Istanbul" and has a Marvel Ethernet controller
The S2937 is not capable using the newer CPU's and is restricted to the dual cores that came out at launch and has a integrated Intel 82571EB Ethernet and a LSI SAS which i am not using.
The hard drives work fine.

The problem is when when i am on the webpage and import the drives and then goto format them the command times out, it is erratic, sometimes it will go through after a few trys or will not at all and stop after 5 times. I have no idea what is wrong other than the common factor between the boards is the same nvidia chipset. I am using the most recent release of nas4free and even tried freenas, and both get write timeouts yet it see's them. I played with the bios settings aswell with no luck. Any idea's?


http://www.tyan.com/Motherboards_S2912-E_S2912WG2NR-E
http://www.tyan.com/Motherboards_S2937_S2937WG2NR

[b0ssman]

when you geth these timeout pls look at the output of dmesg.

there are some problems with freebsd and nvidia sata chipsets.

[TheDataBox]

when you geth these timeout pls look at the output of dmesg.

there are some problems with freebsd and nvidia sata chipsets.

I'm not sure what dmesg is but is this what you want?

kernel: (ada5:ata6:0:0:0): WRITE_DMA. ACB: ca 00 00 00 00 40 00 00 00 00 01 00
kernel: (ada5:ata6:0:0:0): CAM status: Command timeout
kernel: (ada5:ata6:0:0:0): Retrying command

[b0ssman]

yes that looks like the problem.

what you could try is getting a different controller like the m1015

[TheDataBox]

yes that looks like the problem.

what you could try is getting a different controller like the m1015

What about the other board that has the LSI SAS1068E SAS and the Intel 82571EB NIC, i believe it is 8 sata ports total 4 regular sata, and other needing a cable adapter that splits out another 4, only thing is i forfeit the ability to use the newer chips and stuck with the non HE dual core. Considering these factors and costs would it be a fair trade off since all it will do it run 24/7 and have moderate traffic plus the controller you suggested basically cost as much as both MB's? Opinion appreciated.

[b0ssman]

yes give that one a try

[TheDataBox]

yes give that one a try

I been testing it for quite awhile now and it seems to work. The only thing to mention is on the webgui the Diag>INFO>SMART tab will cause the machine to get errors and lockup somewhat other than that it works fine, is it anything to worry about or just leave it alone and i am good?

[RedAntz]

yes give that one a try

I been testing it for quite awhile now and it seems to work. The only thing to mention is on the webgui the Diag>INFO>SMART tab will cause the machine to get errors and lockup somewhat other than that it works fine, is it anything to worry about or just leave it alone and i am good?

Do you see S.M.A.R.T. details after lockup ? Which NAS4Free revision are you using ?

I am wondering if it is because S.M.A.R.T is not getting the right parameters parsed to it.

[TheDataBox]

yes give that one a try

I been testing it for quite awhile now and it seems to work. The only thing to mention is on the webgui the Diag>INFO>SMART tab will cause the machine to get errors and lockup somewhat other than that it works fine, is it anything to worry about or just leave it alone and i am good?

Do you see S.M.A.R.T. details after lockup ? Which NAS4Free revision are you using ?

I am wondering if it is because S.M.A.R.T is not getting the right parameters parsed to it.

No the webpage will timeout and go blank, and then i have to ctrl+alt+del on the Nas system
I am using nas4free most recent release NAS4Free-x64-LiveCD-9.1.0.1.847
Under Disks "ATA" the drive connected to the LSI SAS sata say this
Disk: da0Firmware Rev:
ATA revision: unknown/pre ATA-2
LBA 48: no
Geometry: 0 cyls, 0 heads, 0 spt
Capacity: 0MB
SMART Supported: no
Write Cache Supported: no
APM Supported: no
AAM Supported: no

I can see the HDD's temp but in the smart log it says "smartd[2737]: Device: /dev/da1, failed to read Temperature"

[RedAntz]

Yea, release build doesn't have the controller patch merged from experimental build yet, so S.M.A.R.T. will fail.

You may try the experimental build below that fixed and added a couple of controllers support.
viewtopic.php?f=17&t=4727

It will be great if you can test & report whether it works / not to that thread.

[TheDataBox]

Yea, release build doesn't have the controller patch merged from experimental build yet, so S.M.A.R.T. will fail.

You may try the experimental build below that fixed and added a couple of controllers support.
viewtopic.php?f=17&t=4727

It will be great if you can test & report whether it works / not to that thread.

Sure I will give it a try and see if that get's SMART working.

[TheDataBox]

I just tested it right now and sure enough it did not go on the fritz on that webgui page and no errors in logs and S.M.A.R.T. info was available and seemed to be working fine with the Onboard LSI SAS1068E
I tried just for the heck of it to format the a drive with the NFP3600 chipset but still got timeouts and erratic behavior.


So would it be unwise to use this nas4free release for a permanent solution or should i stick with current one and just wait until it is fixed?

Another thing I have a AMCC 9550SX-12SI RAID card that did the same thing with nas4free, the S.M.A.R.T. would not get passed along and did not get through, would you like me to test that as well?

[RedAntz]

Hi TheDataBox,

Thanks for testing the experimental build. Did you do a 'Clear and Import' when you start using the experimental build ?

Can you be more specific with the 'erratic behavior' ?

I do not know what can cause timeout issues NVIDIA NFP3600, and my Google-fu doesn't give me much either. Perhaps you can ask in FreeBSD forum to see if it works.

To help further troubleshooting , can you PM me your config.xml (make sure to remove your hardisk serial number and your login password stored in the xml)

and please provide me the console output of

Code: Select all

camcontrol devlist -v

and console output of :

Code: Select all

dmesg |grep 'mps\|mpt\|da0\|ahci'

Thank you for your offer to test AMCC 9550SX-12SI RAID card. I will prefer if you wait for the next experimental build that will have a detection patch related to 3ware chipset. I am currently ironing out build issues migrating to FreeBSD 9.2. Hopefully I can release the next experimental build soon.

The difference between my experimental build and release build is that it has latest development patch that is stable enough for wider testing (i do not have all variants of hardware, so i have to rely on user input like yours) before it is merged with main trunk for release build. It is up to you which one you are comfortable to use.

[TheDataBox]

I booted the build you asked me to try and ran it live, i did the 'Clear and Import' and checked the S.M.A.R.T. and everything seemed to function as it should, the only thing i did not do is run an scheduled S.M.A.R.T. self test, i assumed it would have worked and left it at that.

As for using the sata drives connected on the sata ports that run through the NFP3600 chipset, when i go to format the drives it times out a couple times and then says it did go through. It is random it will say it did it on a few trys or try 5 times and just stop, i can create all ZFS stuff and it will seem to work but will get those command timeouts pretty quick. That's what i mean by "erratic behavior' most the time it isn't working but sometimes it does, i played with about every BIOS setting there was to see if that was the cause. It just seems chaotic there is no working or not working, it's more grey and sorta works but ultimately worthless since the whole point of this NAS is a system built upon reliably protecting your data. What is most frustrating is that i think this is the only chipset you can get for socket F as far as i have seen.