Hardware: Juniper QFX5100 for layer 2 vlans. MX240 for layer 3 + routing.
Topology: Layer 3 MX240 -> Layer 2 QFX5100 -> Servers
I have an issue with servers on the same VLAN rarely occurring where TCP connections are being interrupted/lost.
The two servers are on the same VLAN but are on different subnets i.e. 10.10.10.2/24 and 10.10.20.2/24
A workaround I have found and currently implemented for some cases is to split the servers onto different VLANs.
However this is not possible for the current issue I have where it is actually two virtual machines on the same physical server but on different subnets. (I'm aware I can make a virtual adapter on local IPs so it doesn't go back to the switch/router but that's not fixing the issue).
This is not restricted to just one particular VLAN.i.e. vlan 2 with server A and B has an issue with losing remote MySQL connectivityi.e. vlan 3 with server C and D has an issue with server C's mail not being passed off to server D with a "no route to host" error.
When I ping between subnets I see the following "redirect hosts"These example subnets are /29's
64 bytes from **.75.122.252: icmp_seq=2 ttl=63 time=0.406 ms
From **.75.121.121: icmp_seq=2 Redirect Host(New nexthop: **.75.122.252)
From **.75.121.121 icmp_seq=2 Redirect Host(New nexthop: **.75.122.252)64 bytes from **.75.122.252: icmp_seq=3 ttl=63 time=0.405 msFrom **.75.121.121: icmp_seq=3 Redirect Host(New nexthop: **.75.122.252)
From **.75.121.121 icmp_seq=3 Redirect Host(New nexthop: **.75.122.252)
There are no firewall filters in place on the VLANs,
Any suggestions where to begin to look in resolving this would be greatly appreciated.
No comments:
Post a Comment