Hi Team , sorry , i have a basic understanding que...
# random
a
Hi Team , sorry , i have a basic understanding question about the operator native mode vs standalone . whats the exact difference between this two mode . I have went through the documentation and its says now your TM deployment resource management will happen with the help of K8S in Native mode. what does meant by that..? is this means lets say one TM went down and a new TM is needed to be created , now its K8S is bringing up the TM with the help of JM ? and in standalone this was handled by Jm itself and no K8S was involved . .also it talks about a security risk in native mode too what exactly is that .. can someone briefly explain the difference .. thanks in advance
g
The main difference is that in Native mode the JobManager is creating and deleting TaskManager pods.
In standalone mode a predefined number of TMs are created using a Deployment
I highly recommend using the default native mode.
a
Thanks . but even in native mode we are using the same flink deployment with the number of TMs right
Copy code
apiVersion: <http://flink.apache.org/v1beta1|flink.apache.org/v1beta1>
kind: FlinkDeployment
i mean we are defining the same env parallelism and the task slots using which the number of TMs are getting calculated right (env parallelism / task slots) in both native and standalone modes
Also in standalone mode how the TMs are deleted and lets say if one TM is in bad shape how the TM is getting deleted and a new one is created ?