11-05-2013 02:13 PM
We have all of our VNXs, Recoverpoint appliances, and trunks from the 5480s in our blade enclosures terminated in one 5300 per fabric. All of our hosts are running ESXi 5.0U1 with Powerpath/VE. Everything runs at 8G. All switches are running 7.0.2. Qlogic HBAs in HP BL460pG8 servers.
We've seen some storage disconnect messages on or ESX hosts and really strange connectivity problems with the Recoverpoint appliances. EMC support noted that our ports are all set to fillword=0 on the 5300 and I have found fillword at 1 on all of the HP 5480's internal blade/host ports. The trunks ports on the 5480s to the 5300s have fillword set to 3 on the 5480 side and 0 on the 5300 side. We have not changed any of these-these are the settings as I found them.
Should all of these ports both the 5300s and 5480s be set to fillword=3?
11-06-2013 07:26 AM - edited 11-06-2013 07:27 AM
If I remember correctly you want ARB fillwords on 8G ports (which is mode 3?) if they are running 8G.
Appearently you got EMC support in on this.
What does EMC advise you to do regarding to fillword?
11-18-2013 08:06 AM
The fillword in both ends of a ISL should be the same.
If you see counter er_bad_os increase, that means that your fillword is bad and needs to be modified. If you don't see this counter increasing, you dont need to change it.