We experience significantly high rescan times - in the order of 7-10 minutes. This is when host rescans iSCSI software HBA with about 100 dynamic targets configured. I understand the number is a little on the high end but still was trying to make sure nothing stupid is contributing to this.
This is Dell R810 with ESX 4.1 988178. We have intergrated Broadcoms showing up as iSCSI adapters which we don't use. Network wise we have 10G Intel NIC and 1G Broadcom teamed up with explicit failover and 10G being the primary NIC, 1G is backup. We use software iSCSI adapter.
The funny things I've noticed is that Broadcom iSCSI adapters had one host configured as Dynamic send target. While remove/add buttons were grayed out. I found a command to remove that host "vmkiscsi-tool -S -r" and got rid of it. Just to find what appears to be some left overs from VMware debug/QA code added two new static targets to all iSCSI HBAs :
184.108.40.206:3260 iqn.2000-01.com.vmware:test-01 and test.vmware.com:3260 iqn.2000-01.com.vmware:test-01 -- have anyone seen those ?
So I got rid of those as well with :
vmkiscsi-tool -S -r "220.127.116.11:3260 iqn.2000-01.com.vmware:test-01" vmhba37
root@da1-esx235:~# vmkiscsi-tool -S -r "test.vmware.com:3260 iqn.2000-01.com.vmware:test-01" vmhba37
Anyway - apart from being puzzled by appearence of these VMware test targets my question is : how to make sure Broadcom iSCSI HBAs do not interfere with sfotware iSCSI. I did go through manuals and made sure vmnics are not attached to the HBAs and all that. Is there any command to completely remove these HBAs to not let ESX even think trying rescan them ?
Any tips on reducing CPU spikes during iSCSI rescans ?
Thank you all