.
First, I would highly recommend you have a view of my earlier post that summarizes the various (platform independent) optimizations you can implement on an iSCSI fabric. These could be configured upfront, or tried re-actively in case you have any performance or packet loss issues. the post can be found here https://hasanmansur.com/2012/10/14/switch-fabric-setup-for-equalogic-iscsi-powerconnect-switches/
If you see an issue where the MXL switch refuses to boot when placed in the M1000e chassis, be mindful that this may not be a hardware issue. the previous MXL firmware version, specifically version 8.3.16.1, does not support M1000e midplane version 1.0. This however, is resolved in the new revision 8.3.16.2. this is only relevant where you have midplane version 1.0 (check in CMC console via $getsysinfo)
.
Recently, i got the chance to play with the new Force10 MXL blade switches for Dell M1000e Blade Chassis. This is the third installment of the 3 posts i have done covering MXL blade switches. The first one gives an overview of the switches, while the second one gives configuration details for setting these up as a data fabric. The posts can be accessed here:
- Overview, MXL switch for M1000e: https://hasanmansur.com/2012/10/17/force10-mxl-switch-for-m1000e-blade-chassis/
- Data Fabric, M1000e: https://hasanmansur.com/2012/10/17/force10-mxl-configuration-data-fabric-for-m1000e-blade-chassis/
This post gives details on configuring these as an iSCSI fabric, within the M1000e chassis.
For the table below, the values in Column C and D can be adapted, after necessary modifications, to your fabrics.
The Switches were deployed in B Fabric, (as B1 & B2). they were stacked together, but did not uplink to any Core/LAN switch. As such, they were an isolated fabric. The external ports provided connectivity to Equalogic boxes.Since these are to be stacked, the first thing we have to do is set up stacking. The switch only allows the 40G ports to be stacked. For this deployment, the base module ports were used for stacking. QSFP+ built-in ports are set to 40Gb by default while expansion module QSFP+ ports are set to 10Gb ports by default. To change a port from 10Gb mode to 40Gb mode, the switch must be told explicitly that these ports are to be used as stacking interconnects. The following section accomplishes the task. (This assumes the base module ports are to be stacked, but conversion of port mode is being shown only as an example) |
||||
Stack |
||||
FTOS(conf)# |
no stack-unit 0 port 33 portmode quad |
|||
FTOS(conf)# |
no stack-unit 0 port 37 portmode quad |
|||
FTOS(conf)# |
exi |
|||
FTOS# |
write mem |
|||
FTOS# |
reload |
|||
After Reload, a further set of commands to assign port groups as stack group |
||||
FTOS# |
config |
|||
FTOS(conf)# |
stack-unit 0 stack-group 0 |
|||
FTOS(conf)# |
stack-unit 0 stack-group 1 |
|||
FTOS(conf)# |
stack-unit 0 priority 14 |
Assign Manual Priorities for stack roles | ||
FTOS(conf)# |
stack-unit 1 priority 1 |
|||
FTOS(conf)# |
exi |
|||
FTOS# |
write mem |
|||
FTOS# |
reload |
After Reload, verify switches come up as satcked via show system brief and show system stack-ports commands. |
||
Global |
||||
FTOS# |
conf |
since this is an isolated fabric, employ rstp for its superior timers/states, compared to STP (802.1D) based PVST. It is not recommended to disable STP altogether even on isolated fabrics. |
||
FTOS(conf)# |
protocol spanning-tree |
rstp |
||
FTOS(conf-rstp)# |
no disable |
|||
FTOS(conf-rstp)# |
edge-port bpdufilter default |
|||
FTOS(conf-rstp)# |
exi |
|||
Ports |
||||
FTOS(conf)# |
int range tengigabitethernet |
0/1 – 32 , te 1/1 – 32 |
Define internal ports as Layer 2, portfast enabled. Configure MTU at 12000, flowcontrol enabled for both rx and tx. |
|
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
switchport |
|||
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
spanning-tree pvst edge-port |
|||
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
flowcontrol rx on tx on |
|||
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
mtu 12000 |
|||
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
no shut |
|||
FTOS(conf-if-range-te-0/1-32,te-1/1-32)# |
exi |
|||
FTOS(conf)# |
int range tengigabitethernet |
0/43 – 44 , te 1/43 – 44 |
Define EQL ports as Layer 2, portfast enabled. Configure MTU at 12000, flowcontrol enabled. |
|
FTOS(conf-if-range-te-0/43-44,te-1/43-44)# |
spanning-tree pvst edge-port |
|||
FTOS(conf-if-range-te-0/43-44,te-1/43-44)# |
flowcontrol rx on tx on |
|||
FTOS(conf-if-range-te-0/43-44,te-1/43-44)# |
mtu 12000 |
|||
FTOS(conf-if-range-te-0/43-44,te-1/43-44)# |
exi |
|||
VLAN |
||||
FTOS(conf)# |
interface vlan |
50 |
Create a new VLAN for iSCSI. Do not use VLAN 1. |
|
FTOS(conf-if-vl-50)# |
name |
iSCSI VLAN |
||
FTOS(conf-if-vl-50)# |
no shutdown |
|||
Next, assign the port memberships under the VLAN. |
||||
FTOS(conf-if-vl-50)# |
tagged tengigabitethernet |
0/1-32 |
Passing as tagged VLAN to internal blades |
|
FTOS(conf-if-vl-50)# |
untagged tengigabitethernet |
0/43-44 |
Passing the iSCSI VLAN as untagged to the Equalogic boxes, external ports from both switches used. |
|
FTOS(conf-if-vl-50)# |
untagged tengigabitethernet 1/43-44 |
1/43-44 |
||
FTOS(conf-if-vl-50)# |
exi |
|||
Define as many VLANs and port memberships as needed, with the template above. |
||||
OOB |
||||
FTOS(conf)# |
interface ManagementEthernet 0/0 |
Define the OOB Management address. OOB connectivity is provided via the CMC, there is no dedicated port on the external plane. using 172.16.0.10 as an example |
||
FTOS(conf-if-ME-0/0)# |
ip address |
172.16.0.10/24 |
||
FTOS(conf-if-ME-0/0)# |
no shutdown |
|||
FTOS(conf-if-ME-0/0)# |
exi |
|||
FTOS(conf)# |
management route |
0.0.0.0/0 172.16.0.1 |
using next hop 172.16.0.1 as an example. | |
VTY |
||||
FTOS(conf)# |
line vty 0 |
Allow telnet connections to the switch with the default authentication list |
||
FTOS(config-line-vty)# |
login authentication default |
|||
FTOS(config-line-vty)# |
exi |
|||
FTOS(conf)# |
username sample password sample priv 15 |
|||
Hasan Mansur 