WSO2AM version : 1.10.0
I setup API manager after reviewing deployment pattern document.(https://docs.wso2.com/display/CLUSTER44x/API+Manager+Deployment+Patterns)
1 publisher
1 store
1 Gateway manager
3 Gateway worker (Clustering)
2 Load Balancer
2 Key manager (HA)
4 jmeter slaves , 1 jmeter client
and then I did failover test with jmeter. I killed one gateway worker node, during jmeter generating http requests. (gateway worker node 3 -> 2)
I got the result different to that I expected. (little strange) TPS dropped close to zero for 5 seconds, when a gateway worker process shutdown. (when I killed the gateway worker)
Graph - Performance break down when api gateway detached in cluster by killed
I wonder what happen that point? Even though alive gateway workers try to recover something, I didn't think TPS would drop like that.
I personally haven't seen this behavior. Anyway, this can be an issue in either gateway worker cluster or the loadbalancer. To figure it out, you can remove the load balancer and send requests to all 3 gateway workers in parallel. Then kill one worker and see if there are any TPS drop in other workers. If there is, then it's an issue with gateway cluster, if not, the problem is with the loadbalancer.
