AWS messes with networking in unnatural ways. You cannot assume the
laws of normal networking apply - especially when you are involving
their network elements (such as EIP nating).
Are you seeing martian packets in your logs?
I don't know if it applies but I have tried other conventional routing
methods in AWS, only to discover that AWS just doesn't want to behave
normally. Easiest thing is to not fight it, and to defer to AWS's
configs and ways wherever possible. Otherwise there will be pain.
*Brandon Gillespie*
SaaS Operations Director, Kuali
[email protected]
801.682.3444
On 12/4/15 3:35 PM, Matthew Frederico wrote:
Ok, the time has come for my networking "prowess" to take a bow to someone
else ..
*Problem:*
I'm attempting to get both interfaces to route out appropriately to the
internet via its appropriate ENI to extend elastic/ip address limitation.
*Example: *
eth0 -> 172.32.64.0/20 routes out just fine back and forth
eth1 -> 172.32.15.0/20 is able to get pinged okay from other instances but
doesn't respond on the elastic IP ..
*What I have done:*
- I am using iproute2 to add routing rules and route tables
- I have turned on net.ipv4.conf.default.rp_filter = 2
Has anyone had experience dong this or can refer me to someone who can?
*To preempt the razzing i'll likely take:*
- Yes, I have - and "The Google Overlord" told me to eat a 9MM. Oh wait,
that was me.
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
