Hi Community,
We're administrating a multinode Qlik Sense Client-Managed on a IAAS.
Sometimes in the week, we are experiencing some engine issues, for instance, every monday, we've got a peak of user and so engines deployed are not enough to handle the RAM peak.
As we're on a IAAS, we would like to add an engine node at specific time to handle the peak and stop machine when we're back in a normal traffic. We definitely can't have a new machine with 100% uptime to handle RAM peak for 1 morning in the week (cost issues). I Identified two solutions:
1st Solution
The first solution I identified is to just add a node in the cluster and start/stop virtual machine at the peak time. During my test, I saw that TTL ping pollute our Qlik Sense Logs and I would like to avoid it. Moreover, there's a warning in QMC and we would like to avoid it.
2nd Solution
So another solution I find is to add the node dynamically by calling API. Here's the process I developp:
Process at the start of the support Node
1. Start Virtual Machine
2. Call QRS Api : post /servernodeconfiguration/container => to create node (as we already propagate the certificate there's no need to do it again)
3. Call QRS Api : put /virtualproxyconfig/{id} => to adding load balancing
Process at the end of the support Node
1. Delete node via delete /servernodeconfiguration/{id} to delete node (it remove load balacing automatically from virtual proxy)
2. Stop the virtual Machine
I would like to know if there's a better way , if solution 2 is not a risky way in long term support strategy and I'm open to any other idea to handle this kind of problem. I already developped the program and will test it on test cluster, but if you have any feedbacks.
Thank
