Sorry for the duplicate email Eddie. I'm not used to gmail's interface just
yet.

On Thu, Jul 15, 2010 at 3:49 PM, Eddie Wai <w...@broadcom.com> wrote:

> Hello Kai,
>
> FYI, the "bnx2i: iSCSI not supported" message indicates that the port
> associated with eth3 is not licensed to do iSCSI offload.  Echoing Mike's
> question, are you getting this message on all the ethXs that you intend to
> do iSCSI offload?
>
> I get the same message on all of the ethXs, even the ones that we do not
intend to do iSCSI offloading. I was not aware there was a licensing issue
to deal with here. Should I be addressing this with Dell then?

> Eddie
>
> -----Original Message-----
> From: open-iscsi@googlegroups.com [mailto:open-is...@googlegroups.com] On
> Behalf Of Mike Christie
> Sent: Thursday, July 15, 2010 2:05 PM
> To: open-iscsi@googlegroups.com
> Cc: Kai Meyer
> Subject: Re: Centos 5.4 bnx2i iscsi woes
>
> On 07/15/2010 11:57 AM, Kai Meyer wrote:
>
> Just to make sure, you are using something like kernel version
> 2.6.18-194.8.1?
>
> Yes,  2.6.18-194.8.1.el5

>
> Are you getting this no supported error on the other ethXs?
>
Yes, see above response to Eddie Wai

>
>
> When using iscsi_tcp are you using iface binding and with that do you
> still have to do tricks with the routing tables and ARP? Are doing
> binding by iscsi iface hw address/MAC or by the vlan (iscsi iface
> iface.net_ifacename = ethX.Y setting)?
>
When using iscsi_tcp, we are basically leaving all iscsi settings to default
(whatever the iscsi and iscsid services default to.) The routing tricks I
was talking about are implemented in /etc/rc.local, like so:
---
. /etc/sysconfig/network-
scripts/ifcfg-eth2
eval `ipcalc -n -p $IPADDR $NETMASK`
ip route add $NETWORK/$PREFIX dev $DEVICE tab 2
ip rule add from $IPADDR/32 tab 2 priority 200


. /etc/sysconfig/network-scripts/ifcfg-eth3
eval `ipcalc -n -p $IPADDR $NETMASK`
ip route add $NETWORK/$PREFIX dev $DEVICE tab 3
ip rule add from $IPADDR/32 tab 3 priority 300

ip route flush cache
---

and in /etc/sysctl.conf, we added the following two lines:
net.ipv4.conf.all.arp_ignore=1
net.ipv4.conf.all.arp_announce=2

These changes are intended to ensure that inbound traffic to a specific ethX
device is responded to out the same interface it was received on, instead of
the first route in the routing table, as well as to ensure that the ARP
requests are only responded to on the interface that the physical mac
address is on (with out the sysctl rules, eth3 will respond to arps for
eth2's mac).

>From what it sounds like, I should be asking Dell if we are licensed to use
the iSCSI offloader?

-Kai Meyer

-- 
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-is...@googlegroups.com.
To unsubscribe from this group, send email to 
open-iscsi+unsubscr...@googlegroups.com.
For more options, visit this group at 
http://groups.google.com/group/open-iscsi?hl=en.

Reply via email to