fix: JVM_ARGS should be parameterised and configured on jmeter pods #69
Conversation
rahul2393
commented
Apr 17, 2023
•
rahul2393
commented
- Added support of passing JVM_ARGS to pods. Default = -Xms1g -Xmx1g if unset
- Added configuration support for CPU/Memory resource request object. Default = 1000m CPU, 1Gi Memory if unset
- Added configuration support for slave replicas. Default = 2 if unset
|
Setting the JVM_ARGS env in pods get reflected in jmeter process. |
| @@ -22,7 +22,8 @@ | |||
| "nodeLocations": "us-central1-a,us-central1-b,us-central1-c", | |||
| "minCount": 3, | |||
| "maxCount": 3, | |||
| "initialNodeCount": 3 | |||
| "initialNodeCount": 3, | |||
| "jvmArgs": "-Xms1g -Xmx1g" | |||
There was a problem hiding this comment.
We should add a default value for jvmArgs in GKEConfig as -Xms1g -Xmx1g and not mention it in setup.json. Only when we want to override we should pass the value . WDYT?
| @@ -367,19 +372,14 @@ resource "kubernetes_stateful_set" "jmeter-slave" { | |||
| port { | |||
| container_port = 50000 | |||
| } | |||
| resources { | |||
There was a problem hiding this comment.
Why are we removing these resources, this would mean pod can take max availabe resources on a node. Instead we should have keep these values as default and have an option to override these from setup.json .
There was a problem hiding this comment.
Recently we have seen people trying with lesser configuration node in which case if we go with the values as default Machmeter will through Unschedule errors, can you think of any scenario where we don't want the pods to use all the available resources on node?
There was a problem hiding this comment.
In that case what we can do is have a smaller value for requests which can be the minimum requirement for running Jmeter and remove the limits section.
Keeping requests will make sure that when pod is getting scheduled it gets minimum resources to run Jmeter.
One can create 1000 slaves, but when they will try to run , it might fail.
There was a problem hiding this comment.
Make sense, done
