We setup 3 deployment servers behind a LB VIP. We set the VIP in deploymentclient.conf but for those agents we are not seeing them check in. We logged in locally to both a *nix and a Win machine and we can telnet to the VIP on port 8089. The LBer shows that traffic is being distributed between all three. Also, only our original DeploymentServer is showing clients, #2 and #3 do not show any clients at all. All serverclass.conf match and we have serverChecksum set to true in global. All of the app files are identical.
This is new to me also, but after short googling sessio it seems to be working solution.
How about LB’s sticky session setting. Is it on or off? Probably better if nodes call at lest some time to the same backend instead of changing it on every request to different backend node even in same transaction.
who approved this architecture?
for my knoledge, but I could be wrong, it isn't possible to have more DSs behind a LB!
you could have more DSs that manage different clients each one but not using a LB.
If you need more DSs because you have to manage too many clients for only one, you can use more resources (CPUs and RAMs) or more DSs but, as I said, each one has to manage a different list of clients.
In my experience, one DS with normal resources (12 CPUs and 12 GB RAM) managed more than 2000 clients.
Splunk has confirmed themselves that this should work along with multiple posts in this community about the same architecture. I have a plan to roll out an abundance of UFs across our enterprise and was recommended this solution.