Greetings all -
I remember attending a session at the UC about how to setup the Web Adaptor (IIS) to run as a fail over between multiple ArcGIS servers. Essentially what I would like to setup is a routine where the end user clicks on a Flex Viewer link and is served the data from Server A. If after lets say 15 seconds there is no response from Server A (it is down, or has a problem) Server B is automatically loaded. I would only like this to happen at the initial load of the viewer and not for every request. How difficult would it be to set this up, and integrate with Portal on a local Intranet? I will post a question about syncing everything separately as I just want to get the fail over setup before scripting all the syncing together.
Thanks in advance for the help!
You would need a 3rd party load balancer - here's some info from Esri on setting-up a failover:
I'm assuming you are set-up as an active/passive with one license... If you have purchased two server licenses, you may want to look at using the active-active deployment scenario:
I am not very familiar with Portal, but I did come across an Esri doc on that:
Thanks Chris - I believe I want to run the Active-Passive failover. With 10.3, it was my understanding that the license covers a failover installation. I could have sworn you can do it without the load balancer with just the web portal (proxy server).
I will keep researching your presented links and update what I come up with.
For my specific needs, I haven't had a requirement to set-up high availability to make the switch automatically, but have contemplated doing so.
You do get a passive failover license - you'll just need to send an e-mail to firstname.lastname@example.org with your customer info and a request for a passive failover license. As soon as they process, it should show up under My Esri as an available Server license.
Yeah, I have the license, it is just setting it up. High-availability seems like a great concept. That way if one is on vacation or it is the middle of the night, it can switch over and just run until the problem can be addressed. I feel it can help remove some urgency in getting the initial server back up and running.