yes, like any other system implementing TCP/IP that I know the Celerra/VNX creates a network route when you create an interface to point it its network using its own address as the gateway.

That route is necessary for the interface to work and it is always correct (unless you specified the wrong netmask)

It will NOT create a default route

If you cannot server_ping an IP in the same subnet then whats most likely is wrong is:

- cable, wrong port

- switch config

- wrong VLAN

Thanks for the information Rainer, appreciate it.  Will discuss with network guys to try and sort out our issues.

Hello,

We're having the similar issue. Would you please explain more about this?

Do you mean all the upper layers FS would use the interface IP as their gateway and when the traffic touches physical data mover, it just forward the traffic to their relevant VLAN. So it doesn't need to know the VLAN gateway on the switch side.

Thanks,

John

By the way, I'm new to VNX/eNAS. I found it's very hard for us to locate the documentation about how eNAS works,such as administrator guide. Can you help explain how it works a little bit? Such as how physical data mover and virtual data mover work together? the traffic flow, etc. On NetApp, there are vServers, each vServer can have its own gateway and we can specify multiple IP for the default gateway of vServer. On eNAS, I think VDM is similar as vServer, can we specify different gateway for each VDM? Or each physical DM is a big vServer, we can only assign one gateway to it.

Thanks!

the interface route created is a network route for the subnet the interface is serving

Its NOT doing any forwarding to other networks

there is also a default route per data mover that is used in case no other network or host route matches

just like on any other system

since the original poster some 3 years ago never replied back what the problem actually was I dont know what your "similar issue" could be

there is one routing table per data mover that is used by all interfaces

Hi

I would suggest to start with the eNAS documentation

you find find it on support.emc.com by selecting support by product and then eNAS

the CLI reference guide is a good start

Since eNAS code is based on VNX2 most of the VNX2 Series manuals and white papers apply

esp. the ones for VNX networking and Virtual Data movers

On VNX/eNAS there is only one routing table and one default gateway for the DM which is used by all the VDMs

see server_netstat, server_route, server_ping and server_ifconfig

where incoming traffic ends up is determined by attaching an interface - either implicitly by using an interface on a CIFS server or explicitly by attaching it to a VDM for NFS

that has changed on the Unity systems - there each NAS server has its own routing table and default gateway

Thanks Rainer for picking up this for us. We have a new eNAS installed recently, but when we created interfaces which are used for CIFS filers. We found the eNAS would automatically create a route for it, just like kevlee mentioned in the topic.

For example, we create:

interface with ip 192.168.100.43/24 - vlan43, it will automatically create a route to 192.168.100.0/24 via 192.168.100.43.

interface with ip 192.168.101.43/24 - vlan143, it will automatically create a route to 192.168.101.0/24 via 192.168.101.43

interface with ip 192.168.102.43/24 - vlan 243, it will automatically create a route to 192.168.102.0/24 via 192.168.102.43

our DNS/AD is 192.168.150.50 - vlan 50

The physical link between eNAS and Ethernet switch is running with 802.1Q. For each VLAN, it's gateway is on the switch side.

I suppose we should tell eNAS that the gateway IP of each VLAN, but we couldn't as the system automatically created route is already there (192.168.10x/24 via 192.168.10x.43)

Per your advise we define a default route per data mover, in that case, the eNAS knows how to forward the traffic out, but each traffic has a VLAN id tagged with it, the switch/firewall will drop the packet because the VLAN id.

In theory, host x (vlan x) need to talk to host y (vlan y),

the traffic flow is: host x -> gateway-vlan x -> gateway-vlan y -> host y

If we define 192.168.100.1 as the gateway (0.0.0.0/0 192.168.100.1), we have no issues for the communication of 192.168.100.43, but how about 192.168.101.43 and 192.168.102.43? The packets will be dropped because of VLAN id.

Either our deployment/understanding has something wrong, or we should be able to define gateway for each interface, such 192.168.100.1 for vlan 43, 192.168.101.1 for vlan 143, 192.168.102.1 for vlan 243.

I know each physical DM has a default/global CIFS server for antivirus, etc. I think that default route should be for that default/global CIFS server. But how about the other cifs servers on VDMs? do we have a way to define gateway/defaultroute for each of them?

Thanks,

John

see my other reply

there is only ONE default gateway per data mover that is used by all VDMs and CIFS servers

and yes that means that all outgoing traffic not covered by other network routes will use the interface that if local to the detault router IP

Hello Rainer

I have also a similar issue, I try to set a new 10G interface (XXXNASNFS10) on a datamover where we were using 1G interfaces (XXXNASNFS01 for example) until now. The issue is that this new 10G network is on a new VLAN (different than the one used for 1G) and I think I have a problem with the routes as I cannot set the default gateway for this new vlan (vlanid 196 below)

# server_ifconfig server_3 -all

XXXNASNFS01 protocol=IP device=FSN

         inet=172.29.7.69 netmask=255.255.255.0 broadcast=172.29.7.255

        UP, Ethernet, mtu=1500, vlan=7, macaddr=0:60:48:27:fe:6'

XXXNASNFS10 protocol=IP device=LACP10

         inet=10.69.117.62 netmask=255.255.255.192 broadcast=10.69.117.63

        UP, Ethernet, mtu=1500, vlan=196, macaddr=0:60:16:7a:3f:b8

# server_route server_3 -list

server_3 :

default 172.29.0.1 0.0.0.0 REP

net 172.29.7.0 172.29.7.69 255.255.255.0 XXXNASNFS01

net 172.29.0.0 172.29.0.170 255.255.252.0 REP

net 128.221.253.0 128.221.253.3 255.255.255.0 el31

net 128.221.252.0 128.221.252.3 255.255.255.0 el30

net 10.69.117.0 10.69.117.62 255.255.255.192 XXXNASNFS10

host 127.0.0.1 127.0.0.1 255.255.255.255 loop

#server_ping server_3 10.69.117.63 (broadcast address of the new vlan 10G)

server_3 :

Error 6: server_3 : No such device or address

no answer from 10.69.117.63

# server_ping server_3 172.29.7.255 (broadcast address of the existing vlan 1G)

server_3 : 172.29.7.255 is alive, time= 0 ms

If you have any idea on how to set the routes as it creates automatically a route for the new interface but I do not know how to modify the default one to tell it that the gateway for this new vlan is 10.69.117.1 ?? Because for the moment we cannot ping from outside this new IP 10.69.117.62 ...

thank you

just change the default route to an IP of your router on the 10G network

either use Unisphere GUI or server_route

great

Thank you Rainer !

If I change the default route to the router IP on the 10G network will this not affect the current 1G network (172.29.x.x) which is working ?

for clients that are directly on 172.29.x.x it wont make a difference

clients from outside will then use the new gateway - if that works they should be able to cope with it assuming that the routing works