Load Balancer and HA problem

[vmware3222]

Hi everyone,

i configured HA for my nsx EDGE because i want to have 2 edge activ and passv for my load balancing

but whwn i stoppend the seconde first edge VM for calculate the time of the failed connection.

the connection failed and the second VM does'nt work

Do i need configure any thing else ?

Thamk you for yours helps

[cnrz]

‌Is one armed proxy mode, or inline mode used for the Load Balancer?

possible to send "show service highavailability" for both Active and and Standby Edges?

AN internal interface should be chosen by the Edge for HA keepalives, can ESG-1 ping ESG-2 HA interface IP?

[vmware3222]

In fact, i wanted to test HA before configuring load Balancer

i configured just HA and when i shut down my first edge VM . the connection is lost

[ddesmidt]

Looks like the issue here, is more related to the Edge-HA (than LB).

Can you validate the Edge#1 and Edge#2 are well "active-standby" when everything is working fine (in CLI: show service highavailability).

Then unplug the ESX hosting the Edge#1 and validate everything is working fine "...-active".

If there is any issue there, the usual errors are:

. Edge must have at least an interface configured as "Internal" for the HA to work

. ESX2 (hosting the Edge#2) VXLAN-transport interface can not talk to the other ESXi VXLAN transport interface (try ping ++netstack=vxlan x.x.x.x [-d -s 1570]).

. Assuming the Edge uplink interface is on a VLAN, ESX2 (hosting the Edge#2) doesn't have the port group well configured for Edge uplink

Thanks,
Dimitri

[vmware3222]

Thanks for your response

yes the problem is HA not load balancer

I configured HA exactely like this link

Virtualization The Future: NSX Deepdive Part 11 - Configuring NSX Edge High Availability

but i don't know what i can show if the second edge on the standby state

[cnrz]

As in the example, what are the IP and Subnet Mask  given for Management IP Addresses? Is it possible to ping to EDGE-2 Management IP from EDGE-1? Also same command is used for verification for HA in the example link.

show service highavaillability command may be entered after ssh to the Edge, or from the Console of the Edge VM.

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=21265...

As pointed on the previous post: If the output of the below command is different than below, then there may be a problem between communication of  HighAvailability interfaces. Since they are most likely on the same Logical Switch, troubleshooting VTEP, MAC or ARP Tables may be necessary. Controller cluster and ESX Hosts  needs to be communicating. Simple check may be communication channel health-check:

http://www.virtually-limitless.com/nsx/nsx-6-2-communication-channel-health-check/

EDGE-1-0> show service highavailability

Highavailability Status: running --> this needs to be running on both EDGE-1 and EDGE-2
Highavailability Unit Name: edge-1-0
Highavailability Unit State: active --> this needs to be active for EDGE-1, and standby for EDGE-2 (or active for EDGE-2 and standby for EDGE-1, one of them active, the other is standby)
Highavailability Interface(s): vNic_1 --> Is this and Internal (not Uplink) interface?
Unit Poll Policy:
  Frequency: 3 seconds
  Deadtime: 15 seconds
  Stateful Sync-up Time: 10 seconds
Highavailability Healthcheck Status: --> Is this host and Peer host healthcheck status good?
  Peer host [edge-1-1 ]: good --> During the failover test Peer Host becomes unreachable
  This host [edge-1-0 ]: good
Highavailability Stateful Logical Status:
  File-Sync running
  Connection-Sync running
      xmit xerr rcv rerr
      21612 0 13920 0