Need help to troubleshoot ip routing between two switches

[GaganGang]

I am facing a outlandish difficulty on my network from remote offices. I have 2 inputs linked via service provider WAN link. For example, GW1 and GW2. That is biologically situated in two dissimilar places. at the moment there are two LAN switches (no IP routing is enabled), for example SW1 and SW2 are connected to gw1.on the other hand, only a switch, for example SW3 is connected to GW2.in SW1, two computers, for example h1 - h2 and 192.168.1.1 - 192.168.1.2 in sw2, two computers, for example h3 - h4 and 192.168.1.3 - 192.168.1.4 subnet / 24. The 4 host are under the same vlan 300.On the other hand, SW3 is connected to GW2.There are 2 servers connected to this SW3, for example, S1 and S2.S1 - 192.168.4.1 and S2 - 192.168.4.2 subnet / 17.Now the question is:

• h1 is able to ping S1 S2,
• h2 is able to ping S1 S2,
• H3 is able to ping S1 S2,
• h4 is able to ping S1 S2,

Having said that, the first troubleshootin has done is connecting to h2 sw2, which does not help. So I cannot actually suspect sw1/sw2 configuration difficulty. Only are identical. I checked all the L2/L3 switching and routing level, nevertheless at a halt could discover a number ofclues to this outlandish behavior. The destination is / 24 subnet and the 4 teams are on the same subnet and the same VLAN. Some ideas for problem solving are highly appreciated.

[Amie]

To ping a host to a further device, first send an ARP request with the expect that the device you are trying to react to ping the host to notify its MAC address. If for a number of cause the device is not within the same broadcast domain or other network device responds to the ARP request in its place of this may be why you cannot ping to a. Separate VLAN broadcast domains as well as separate switches. As one would probably take diagram much closer to finding what is wrong. When you know how things are physically / logically connected then you can begin scanning the layer 2 / 3 and the layer above.

[HiSpeed]

in gw1: 
ip prefix-list 192.168.0.0/17 allows ENTRY next 10 
ip prefix-list 192.168.1.0/24 allows OUTGOING next 15 
and gw2: 
ip prefix-list 192.168.0.0/17 allows ENTRY next 10 
ip prefix-list 192.168.1.0/22 allows OUTGOING next 35

Distribution lists input and output interfaces are applied to serial GWS. Also if I do 'sh ip route "I am receiving flow paths from the right direction. I use "eigrp" as my routing protocol redistribution.

[BiGMaCHiNe]

Have a prefix of the list blocking the ARP request will not prevent the system to try arp because, for starters.192.168.0.0/17 includes all addresses from 192.168.0.0 - 192,168,127,255 - consequently, if the VLAN 300 hosts attempt to converse to 192.168.1.3 - which will mean with the intention of the host is local to the subnet and send an ARP request out. Never send it to the default gateway. These overlapping subnets. With the exception of the implementation of NAT, this breaks the logic of IP. No matter what is configured on the devices.

[Connect_Me]

I respect people certified (or people for that matter), but this is bull. Suitable subnets is the second thing you need to know in the nets with the OSI layers, however, we are advocating here a design that breaks with the logic, and I do not see why it fails. Somebody who comes to this design in my work would be fired. Excellent fortune in deductibles, I disappear this concern (and I do not want to feed the trolls).

[Sabastiaan]

gw1 its gives me the right direction only - 192.168.0.0/17 learned from the serial interface via EIGRP. 
for example, gw2: 
your internal network 
gw2 # sh ip route 192.168.4.1 
Routing entry for 192.168.4.0/23 
Known via "eigrp xxx", the distance 90, 3072 meter, internal type 
EIGRP redistribute through xxx

its external network .... 
gw2 # sh ip route 192.168.1.1 
Routing entry for 192.168.0.0/17 
Known via "eigrp xxx", distance 90, metric 9216, type internal 
EIGRP redistribute through xxx