[jboss-jira] [JBoss JIRA] (WFLY-9835) Performance issues in WildFly11 as my 90% for jmeter run gets increased to 2x for same set of run as it was in Jboss.

nilesh ratta (JIRA) issues at jboss.org
Thu Feb 15 07:28:02 EST 2018


    [ https://issues.jboss.org/browse/WFLY-9835?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13533771#comment-13533771 ] 

nilesh ratta commented on WFLY-9835:
------------------------------------

Hi Jason . Because of performance issues , i have been stucked , As i  am not able to roll out WildFly11 in production environment.
So kindly provide your inputs as i have shared all relevant informations.
I have also shared Garbage Collector logs and heap dump snapshots.

> Performance issues in WildFly11 as my 90% for jmeter run gets increased to 2x for same set of run as it was in Jboss.
> ---------------------------------------------------------------------------------------------------------------------
>
>                 Key: WFLY-9835
>                 URL: https://issues.jboss.org/browse/WFLY-9835
>             Project: WildFly
>          Issue Type: Bug
>            Reporter: nilesh ratta
>            Assignee: Jason Greene
>         Attachments: J-img3.png, J_img1.png, J_img2.png, Jboss_GClog.png, W_img1.png, W_img2.png, W_img3.png, WildFly11_GCLog.png
>
>
> I have upgraded my application from Jboss6 with java 6 to WildFly11 with java 8 in dev environment.
> Note :
> I am facing some performance issues in WildFly11 as my 90% gets increased for same set of run as it was in Jboss.
> *And minor GC gets increased to 7 times as it was in jboss.*
> And there is no application level code changes in wildfly11.
> *Please find attached files for GC log and Heap dump related inputs.*
> Inputs :
> JVM parameters : Parallel GC with 4 threads , -Xmx6144m Xmx6144m  -XX:NewRatio=1 -XX:ParallelGCThreads=4 -XX:MetaspaceSize=96M -XX:MaxMetaspaceSize=1024m
> Parameters are same in Jboss 6 and WildFly11.
> Jmeter result overview for 100 users in 15 mnts with  250 sec rampup time.
> As all of our app server is on AWS with same configuration.
> 1. Jboss 6 results :
> Label      Count	  Average    90%_line	Min     Max	   Error%    ThroughPut
> TOTAL	   48664	  167	     418	        0	   30374       0	           53.80															
> 2. WildFLy11 results :
> Label	   Count	   Average	  90%_line	 Min       Max	    Error%	  ThroughPut
> TOTAL	   54775	   287	          887	         0	    14286	     0	          60.672
> Note : As my count nd throughput increases but my* 90% is > 2x * for WildFly11.
> Kindly find the attached files for Heap dump as follows with following names  in .png format:-
> For WildFly 11 :
> 1. W_img1.png :-    
>    WildFly11_DominatorTree :  io.netty.buffer.PoolChunk   ?????             
> 2. W_img2.png 
>    WildFly_Histogram       :  Class Name| Objects | Shallow Heap |  Retained Heap
> 					 byte[]    |  53,925 |  163,144,832 | >= 163,144,832
> 3. W_img3.png
> 	WildFly_Immediate Dominators for byte [] :
> For Jboss -6
>   
> 1. J_img1.png :-
>     Jboss6_DominatorTree   : as there is no  io.netty.buffer.PoolChunk  .....
> 2. J_img2.png :-
>    Jboss6_Histogram 
> 3. J_img3.png :
>    Jboss6_Immediate Dominators for byte [] :
>    
> Kindly find GC log images for wildFly 11 and jboss 6 .for 15 mnts load run .
>                         Wildfly 11               Jboss 6
> 				
> Total GC                1162		              174
> Total GC Time           >21 sec                  6 sec
> Total Reclaimed Bytes    3.32 tb                  480 gb
>  



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)


More information about the jboss-jira mailing list