lhrslsshahi
asked on
AWS - HAProxy- KeepAlived
Hello experts,
We use HA Proxy on Centos 7 so that it load balances incoming web traffic as well as internal requests between the web and application layers.
For the failover we are using Keepalived VIP however in AWS multicasting isnt't supported but in our VMware environment it works just fine.
Is there anyway of using HA Proxy and Keepalived with VIPs rather than unicast and source ip in AWS.
Currently I am using unicast and source ips in Keepalived. What I have done is create 2 DNS A records instead of using a VIP.
For the incoming web traffic I am using EIP instead of VIP which is fine. I want to address the internal loadbalancing requests.
Haproxy1 Haproxy2
| |
Web1 Web2
| |
App1 App2
| |
DB DB
We use HA Proxy on Centos 7 so that it load balances incoming web traffic as well as internal requests between the web and application layers.
For the failover we are using Keepalived VIP however in AWS multicasting isnt't supported but in our VMware environment it works just fine.
Is there anyway of using HA Proxy and Keepalived with VIPs rather than unicast and source ip in AWS.
Currently I am using unicast and source ips in Keepalived. What I have done is create 2 DNS A records instead of using a VIP.
For the incoming web traffic I am using EIP instead of VIP which is fine. I want to address the internal loadbalancing requests.
Haproxy1 Haproxy2
| |
Web1 Web2
| |
App1 App2
| |
DB DB
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Yes, you can use a Route53 health check. I usually use an endpoint check, but that wouldn't work for internal instances.
You can do a health check based on a CloudWatch metric (i.e. trigger if CloudWatch detects host is down). This is a bit slower to failover though.
You can do a health check based on a CloudWatch metric (i.e. trigger if CloudWatch detects host is down). This is a bit slower to failover though.
ASKER
Will let you know how I get on. Thanks
ASKER
Thanks Phil.
ASKER
What about AWS DNS failover and health checks so if the primary fails and it will go to the sencondary automatically rather than me manually changing the CNAME to the second A record when the failure occurs?