vmware esxi 5.5 and adaptec 6805 raid, host hangs on high i/o load

[kira_belka]

Dear  Mr/Mrs!

I faced with a trouble.

Well I have  2x supermicro 1U server ( 2 x Xeon 2620, 128Gb RAM ) and  adaptec 6805 controller within raid 6 (6 disks(2Tb) in raid, 1 as hotspare).

I installed ESXi 5.5 build 1746974 (customized using ESXi-Customizer-v2.7.2) on both hosts.

adaptec driver included in install has version vmware-esxi-drivers-scsi-aacraid-550.5.2.1.40301.-1.5.5.1331820.x86_64.

And I got a system crush when I 'd tryed to copy big file(more than 30Gb) from one place in datastore to another(on one host).

Also hosts hangs trough unexpected intervals of time (on next day.. three day).

Hosts has tested and verified hardware.

I attached error in logs.

Looking forward to hearing from you.

Thx for any possible help.

Youth Faithfully,

Kira

[Punisher713]

Hello Kira,

I've been experimenting the same problem with a 5805 adaptec controller. Contacted Adaptec for support, they told me from the logs my hard disks were incompatible with the controller. So I spent another 1000$ and bought new disks, listed as compatible, still the same problem persists. Right now i'm trying to install windows server 2012 R2 straight on my Raid array to see if I get further than with ESXi 5.5. I have ordered a brand new LSI MegaRaid 9271-8i because I'm tired of fighting with the adaptec controller.

Just to let you know what i've already done without results so you don't wate your time:

• Updated firmware
• used original driver in esxi 5.5 installation media
• tried third party driver
• tried driver from adaptec's website
• enabled/disabled cache on controller
• contacted adaptec for support
• tried to get help with the LED error code I get on the controller; no answers online, Adaptec said they don't use LED error codes for diagnostics anymore and asked that I send them logs with ARCCONF utility. found nothing there
• Re-created the array multiple times
• tried with desktop drives and Constellation enterprise-grade drives
• Tried creating a vmdk of 1TB on array: freezes instantly
• creating a VMDK of 40gb was succesful. After creation, expanded to 80gb. Then expanded to 120, 160 and so on until I reached close to 400gb.Tried to write to that drive, crashed after a minute or so.
• I didn't have any trouble reading what I wrote to the drive, but wirting to it would make the system Hang
• Installed Windows server 2012 R2 straight on the array: Install was successful, can't get into windows. I keep getting a bluescreen with error "MACHINE_CHECK_EXCEPTION". system hangs there and I have to force a reboot.

from what I read on other forums and threads, it seems we have defective controllers. If you have any other ideas or found a solution to the problem, please let me know.

Regards,

Marc-Andre.

[JBabalan]

Try to change Maximum Payload value to 256 Bytes in BIOS under PCI-e Configuration

JB

[samdeng]

Hi, i faced the same problem. update firemware, get adaptec support..etc....

When i disable VT-D in BIOS , the VM GUEST OS has not hang for a long time.(about more than 1 month+), but today i face the problem again. ....

I guest , the vmware esxi  lost the storage device .

VMWARE ESXI 5.5 1623387.

the next : i will check the IOMM and Maxinum Payload value in BIOS---> PCI-E.

[Punisher713]

Thank you all for your replies.

In my case, disabling VT-D did not change anything. I didn't try max pay load yet...

I ended up buying a LSI MegaRaid 9271-8i and everything worked plug and play! After weeks fighting agaist my 5805 i needed to get the server running.

I will do more tests with the 5805 on another machine sometime. I need to see if the card is defective before i try to resell it.

[krumedia]

Hi,

does the 9271-8i work with the free(!) version of ESXi 5.5 ???

I also need to replace the 6805 due to the same trouble...

[samdeng]

It's not working in 6805 in my condition. now i has replaced the 6805 to LSI 9240(9270). also downgrade to ESXi 5.1. it's very nice.

[cneulieb]

Figured I would respond as I have been looking for a solution to this for a few months.

http://ask.adaptec.com/app/answers/detail/a_id/17400/~/vmware-esxi-5.5%3A-unresponsive-system-and-sc...

I haven't been able to get it implemented yet, but it looks like the problem may be addressed.