VMware Cloud Community
rhchodur
Contributor
Contributor
Jump to solution

HA and DNS

I did some research but could not come up with a good answer to how this works.

The recommendation is to have two Service Consoles for HA so there is two paths and less of a chance to have a false read.

HA depends on DNS, so if you setup two service consoles with two different IPs then add these IPs as A records to DNS with the same name you will rely on round robin to return a different IP for each query.

Ok so does HA make a full DNS query before each check for a server to be up so the IP and test would alternate between IPs/adapters each time, or does it make only a query when the cached resolution times out. If the second is happening then what good is the second service console when you would need to flush DNS resolution cache for it to know if the server is up and you rely on DNS resolution to what IP it tests on based on where the round robin is at in query requests at any one point in time.

Anyone dug more into this?

For now we are using two nic teaming on one vswitch with the SC using one nic primary and VMotion using the other as primary, each failing over to the other. Only one service console.

0 Kudos
1 Solution

Accepted Solutions
ZMkenzie
Enthusiast
Enthusiast
Jump to solution

Ha does a "hostname -s" for each known esx server and try to ping that host (the short version of your esx name), it does a query everytime it's configured i suppose since linux does not have any dns caching system. So, just when you activate ha or restart your server.

Anyway it will ping that address first, then check the default gateway for each of your service consoles, then, it will check any das.isolationaddress set in advanced properties.

So, you don't need to have 2 service consoles i think, just set up your service console on a vswitch with more than one interface and then add some das.insolationaddress on ha options using pingable ips in the same network.

Hope this helps.

View solution in original post

0 Kudos
3 Replies
ZMkenzie
Enthusiast
Enthusiast
Jump to solution

Ha does a "hostname -s" for each known esx server and try to ping that host (the short version of your esx name), it does a query everytime it's configured i suppose since linux does not have any dns caching system. So, just when you activate ha or restart your server.

Anyway it will ping that address first, then check the default gateway for each of your service consoles, then, it will check any das.isolationaddress set in advanced properties.

So, you don't need to have 2 service consoles i think, just set up your service console on a vswitch with more than one interface and then add some das.insolationaddress on ha options using pingable ips in the same network.

Hope this helps.

0 Kudos
rhchodur
Contributor
Contributor
Jump to solution

Thanks ZMkenzie - the das.isolationaddress piece was what I was missing to move forward.

0 Kudos
HNNY5067
Contributor
Contributor
Jump to solution

did you ever implement the das.isolationaddress? If yes, how did that work out?

0 Kudos