DSS6.0up10 Replication problem I/O errors

[eppo78]

Hey all,

I'm having serious issues with replicating 1TB FC volumes to my destination DSS.
I'm running DSS6.0up10.8101.3719 64bit on the following hardware;

SuperMicro X7DWN+ Motherboard
64GB Memory
Qlogic QLE2462
Areca ARC-1680 4096MB cache Firmware V1.47 (Raidcontroller)
24 SEAGATE ST3500620SS (SAS 500GB 7200RPM)
1 10GB Areca Volume (DSS6 OS)
1 10990GB Areca Volume (DSS6 DATA)


I have created a FC volume of 1024GB with replication and initialization enabled and gave the volume LUN 0.

I have severely tested the performance with IOMeter 1TB datastores in VMware ESX4 and it is outstanding . DO NOT use a 2TB datastore in VMware, you WILL get I/O errors!

However when I create a replication task with a 1GB nic set at 60MB for the 1TB volume to a mirror server, the source server hangs after a period of time.

I get the following errors from mail and the console:
mail: kernel:journal commit I/O error
console: arcmsr0: abort device command of scsi id = 0 lun = 1

You would think this is a raidcontroller issue but I replaced the controller for a new one including the memory and because I severely abjused the system for 3 days continuously using IOMeter, I think it would have crashed then if it was a hardware issue.

So this issue only occurs when the replication task is running.

Has anyone had the same problem?
Is it possible for anyone running DSS6 with a Areca 1680 to test if they can reproduce the same issue?

Your Help Would Be Greatly Appreciated!

(I already send multiple mails to support about this issue, but without response so far)

[eppo78]

Oh yeah... Because the server frooze up I had to do a hard powercycle.

The result was a corrupted systemvolume that I was not able to repair, all my data was lost

[symm]

try using LUN 1
in esx 3.5 it did not like lun 0

Are you replicating through the Qlogic QLE2462?
if not what type of nic are you using

[eppo78]

try using LUN 1
in esx 3.5 it did not like lun 0

Are you replicating through the Qlogic QLE2462?
if not what type of nic are you using

I don't think it is ESX related as I have the same problem with a empty volume without a LUN assigned to it.

As far as I know you can only replicate through a nic ( Intel 82575EB internal)
Thanks for you reply though.

Eddo

[eppo78]

The same issue occurs while replicating a 500GB volume...

[symm]

Did you try sending the logs to support?

did the initilization finish before you started the replication ?


BTW it does not seem like a vmware issue, but here is a link to the post concerning LUN 0

http://kb.vmware.com/selfservice/mic...8601&stateId=0 0 44972208