Unsolved
This post is more than 5 years old
6 Posts
0
853
eNAS static route issue
Good Afternoon,
We just have our eNAS system deployed, while doing the setup, we found a route issue.
The eNAS is running on VMAX ALL FLASH, and it's connecting to Cisco Ethernet switch through a 802.1Q trunk link.
We found when we create an interface on the eNAS physical data mover, it will create a static route by itself automatically.
For example, we create an interface with an IP 192.168.10.50/24, vlan id is 50, it will create a static route like:
192.168.10.0/24 via 192.168.10.50
My question is what is that route for? Because we want to have a static route like 192.168.10.0/24 via 192.168.10.1 to route traffic out.
192.168.10.1 is the real gateway for vlan 50 and without define it on eNAS, no traffic is going out.
Because the system static route exist, we can't do that (adding net route 192.168.10.0/24 via 192.168.10.1).
In that case, if we create 10 interfaces on eNAS, they could not know their own gateways. The support tell us we need to leverage system default gateway. I don't see how it works, When each packet is sending out, it has a VLAN ID attached, only its own gateway can remove that VLAN ID when forward the packet to another VLAN. So even we have a gateway defined, most packets would end with being dropped.
I'm new to VNX/eNAS, any hints would be helpful. Thanks in advance!
jsha
Rainer_EMC
8.6K Posts
0
August 14th, 2018 01:00
thats the normal network route that every IP system creates
If you create an interface with 192.168.10.50/24 it means that traffic to all IP's for that subnet 192.168.10.0 - 192.158.10.255 will be sent out through the local interface 192.168.10.50
traffic to IP's outside of that subnet will be sent to the default router
Thats the way IP routing works - if the dst is local then the packets get put in the wire with the dst MAC of the local dst - if its not then it gets sent to the dst MAC of a router
You cant define a subnet as local but then want it to be sent out to a router - doesnt make sense to me
I would suggest to discuss with your network department
jsha1
6 Posts
0
August 14th, 2018 09:00
Thanks Rainer. It turned out to be we should NOT try to use the 'Run_Ping_Test' from eNAS to ping an external IP from an interface. It looks eNAS doesn't rely on it when it's working. that's a wrong verification method we're told.
If we try, the issue is still there, the eNAS will put the traffic to default route, then the packet will be dropped as VLAN ID mismatch.
My understanding,
eNAS contains CIFS server, the interface IP is linked to the CIFS server, such as 192.168.10.50. the CIFS server doesn't have its own gateway set (for example, 192.168.10.1). Like a Windows server, it only has its IP set, but no gateway defined. Gateway is used to forward traffic to another subnet. eNAS does have the gateway connected - it's on the other side of 802.1Q link between eNAS and switch.
It does make sense to define a default route for local interface/subnet. the traffic flow would be:
192.168.10.50 -> 192.168.10.1 -> 192.168.20.1 (DC's gateway) -> 192.168.20.50 (DC)
I still don't know when it needs to start communication to outside, such as DNS, authentication/DC, what source address to be used. I believe it's not 192.168.10.50. It's like there is a proxy there for all interfaces to forward the traffic to default route. It's like on 802.1Q trunk, we have vlan 10, 20, 30, 40, 50, each vlan's gateway is resided on switch side, but we will only use vlan 30 (192.168.30.1) as the default route for every thing.
I also don't understand your saying 'traffic to all IP's for that subnet 192.168.10.0 - 192.158.10.255 will be sent out through the local interface 192.168.10.50'.
192.168.10.50 is a host IP for CIFS server, it's supposed to receive nothing but destined to 192.168.10.50. What's the meaning of the route -- 192.168.10.0/24 via 192.168.10.50
Rainer_EMC
8.6K Posts
0
August 16th, 2018 15:00
to verify correct routing use server_ping
do display the routing table use server_netstat -r
all VDMs use that global routing table on the physical data mover - thats where the default route is set using server_route
for outgoing traffic that is not as a reply to a NFS/CIFS request it depends on the context
typically is the the IP of the first (I think) interface attached to the CIFS server
I would suggest that you ask for a local specialist from your area to talk this through with you in person.
Or open a service request with network traces that show your problem
It seems to be too difficult to troubleshoot over forum posts
Rainer_EMC
8.6K Posts
0
August 16th, 2018 15:00
not sure if we are talking about the same thing.
If we (as in the CIFS server) need to send a paket to any host in that subnet we will hand it over to the data mover interface that has been configured with the IP of 192.168.10.50 - for example cge0
This interface will then put it directly onto the network since its a local subnet