Timeouts with HP NC522sfp+ (QLogic NetXen) 10GbE cards and ESXi 4.1
We have a pile of brand new HP DL380 G7s with NC522spf+ cards. We already dealt with the recent PSOD problem when combining firmware 4.0.520 and up with the as-shipped driver in ESXi 4.1. We are running firmware 4.0.526 with the latest VMware nx driver, 4.0.570 (164009).
The cards work fine initially. We can pass management network vmk and iSCSI initiator traffic, and can discovery LUNs fine. Once we try to mount a VMFS partition, things go belly up. The host can't read the VMFS signature on existing LUNs and can't format new ones without the NICs timing out, and eventually dropping all connectivity from the network.
Has anyone gotten the NC522sfp+ to work reliably with ESXi 4.1? We are looking for the magic firmware and driver combination to resolve the timeout issues.
We are working other channels, and may have to just swap these for the Emulex based NC550sfp.
While we did have PSOD issues with the NC522spf, including a disconnect at one point (possibly due to cooling), our iSCSI issue turned out to be self inflicted. The problem was a hard set MTU of 1500 on the interfaces on the Juniper EX 4500 switches we are using. The correct default is 1514 for Junipers due to how they calculate MTU. This caused a classic MTU black hole when sending big packets between the Juniper and Cisco switches (which the NetApp heads are connected to.)
If anyone has comments on stable firmware and driver combos, it would still be appreciated.