I would like to know from those using QLA 2342 or 2642 qlogic hba card and that have researched for recommended setting, what values you are using for those options:
Adapter Setting
Fibre Channel Tape Support : Disabled
Advanced Setting
Execution Throttle : I have the value 16
Login Retry Count : I have the value 8
Port Down Retry Count : I have the value 14
Link Down Timeout : I have the value 30
Operation Mode : I have the value 0
Interrupt Delay Timer : I have the value 0
Enabled Interrupt "2462" : I have the value No
Thanks
Richard
We kicked the tires on this basic issue for some time... the short answer, defaults work fine. We had a couple of 'know it alls' demand we test at 64 versus 32, so we did. Never mind VMware and QLogic say, never change this unless there is a specific reason. For example, in a typical virtualization environment, read/writes are completely random in reference to total IO and total time for IO transport. However, if the majority of your VMs are doing something odd, like many many writes, taking longer or many many reads that are fast and you guess OSes are very sensitive to timing, there might, just might me a reason for messing around with the HBA queue length in one direction or the other. We found that increasing the timeout value in the VMs themselves was a better solution over all, then tweaking ESX its-self. When an ESX host is hurting, do you really want that much stuff queued up for commit to disk? No. so increasing the length to 64, and depending on the type of storage solution you have, it could get real ugly.
Anyone!
Using these settings on qla2460 and all works fine here. Some people change the Qlength from 32 (standard) to 64, but IMHO this is not needed (under normal circumstances).
Have you any problems there?
We kicked the tires on this basic issue for some time... the short answer, defaults work fine. We had a couple of 'know it alls' demand we test at 64 versus 32, so we did. Never mind VMware and QLogic say, never change this unless there is a specific reason. For example, in a typical virtualization environment, read/writes are completely random in reference to total IO and total time for IO transport. However, if the majority of your VMs are doing something odd, like many many writes, taking longer or many many reads that are fast and you guess OSes are very sensitive to timing, there might, just might me a reason for messing around with the HBA queue length in one direction or the other. We found that increasing the timeout value in the VMs themselves was a better solution over all, then tweaking ESX its-self. When an ESX host is hurting, do you really want that much stuff queued up for commit to disk? No. so increasing the length to 64, and depending on the type of storage solution you have, it could get real ugly.
The reason that I am asking it is that for a over a year, we have been using the values I mentionned. But two weeks ago, when I updated the firmware on my QLA2462, the execution throttle that was at value 16, changed to 255. It sounds like the default value is now 255.
Richard
Here are my settings. All other settings set to default.
Hardware is HP BL460c, attached to an EMC Symmetrix.
Hard Loop ID: 125
Fibre Channel Tape Support Disabled
Data Rate 3
Execution Throttle 256
LUNs per Target 256
Port Down Retry Count 30
Link Down Timeout 30
The following settings are required for ESX Server operations on the Symmetrix networked storage system:
Common serial number (C)
Auto negotiation (EAN) enabled
Fibrepath enabled on this port (VCM)
SCSI 3 (SC3)
Unique world wide name (UWN)
SPC 2 (Decal) (SPC2) SPC 2 flag set
Hope that helps...
Rgds,
Didier
HBA settings should be configured for what your storage vendor specifies. Many times the vendors release "bios versions" specific (tuned) for their storage.
On the other hand, most of the time the defaults will be fine. The fine tuning will help with things like fail over and resets.
I haven't found yet a configuration that solve the reset issue. IBM says that to follow vmware recommendation for the DS4000. So we did that but still not failling over when reseting a controler. Now I am trying to get IBM in a conference call with a VMWare engineer.
Have you got into a call and got the issue solved?
My HBA settings for port down retry count is 15 and i am too facing the not failing over issue as you.
Can you please let me know how you solved this issue?
VBDINO/SSATISH,
Have either of you been able to get a consensis from IBM/VMware on the DS4000 issue? I've run into the exact same situation!
Thanks