I have an install of glassfish 4 on centOs 6.7 hosted at 1 & 1.
The config is the following :
1 VCPU
1GO RAM
40GO SSD
My max numproc is 160 and whenever I run glassfish and do some operations, I reach the maximum and the server crashes.
I've followed oracle performance tunning guide but not much success with it.
Do I need a server with a config more powerful ?
Thanks in advance
Related
We have two wildfly 16 server running on Linux. First with JDK 11.0.2, second with JDK 8.
Wildfly 1 has a remote outbound connection to wildfly 2 which is used for HTTP-remoting. This is necessary because it has to run with Java 8 32 bit.
When we perform a load test after 100.000 requests from wildfly 1 to wildfly 2 response time increases steadily.
A heap dump analysis of wildfly 2 using MAT gives us some information about the problem. The heap dump shows a lot of ‘io.netty.buffer.Poolchunks’ that use about 73% of the memory.
Seems the inbound buffers won't be cleaned properly.
Wildfly 2 does not recover when the load stops.
Is there any workaround or setting to avoid this?
I am working with a ubuntu web server, where I have tomcat 8 and running a web application on tomcat. Recently I have faced some problem with cpu uses 100%. When I restart the tomcat server it is running good, but after one day or a few hour again same problem arise ( 100% cpu use) and that problem make my site slow. When I see the process list with htop command then i see so many process like
/opt/java8/jre/bin/java -Djava.util.logging.config.file=/opt/tomcat8/conf/logging.properties -Djava.util.logging.manager=org.apache.juli.ClassLoader
The website is a live eCommerce website and average user in every moment is around 100.
The server is a aws ec2 server where a tomcat and a java application inside the tomcat is running. The database (MySql) is in an another aws RDS server.
What can i do in this situation? Please help.
Here is the gist for my test
I am trying to test the performance of neo4j writes (embedded and rest) using jlbh. Unfortunately i am not able to get anything beyond 50 nodes insertion per second (((
I am hoping that this is something to do with how i am configuring the multiple performance tuning nobs.
Config details:
the jvm is oracle jdk1.8.0_77
the server ulimit is set to 40000
the server is an intel Ubuntu 14.04 server with 24gigs of ram.
uname is (Linux ymo-dt 3.19.0-58-generic #64~14.04.1-Ubuntu SMP Fri Mar 18 19:05:43 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux)
both the client and the server are running on the same machine (obviously since it is embedded)
i figured once i know how to tune the embeded version i can go ahead and try the rest api with both client and server running separately
I am trying to follow this so question in the tuning
As the gist (gradle part) says the version of neo4j i am testing is 3.0.1 from amven
I am passing Xmx and xms as both set to 4096m as per the gradle file
I cant for the life of me understand how that person got 7070 nodes insertion per second.
My questions :
What am i doing wrong ?
What do i need to add to the perfTuning section to make this reach 7000 nodes insertion per second ?
I am hoping that i am not going to need to batch the inserts at all. So the actual question is what is the maximum inserts i can achieve without resorting to batching ? or how ami supposed to do it with one tx per insertion ?
Once i move to the rest api what is the most performance tuned api i can use to access the server remotely ? I tried bolt and i am also not getting more than 38 nodes inserted per second.
Edit:
Seems like ext4 is to blame here on linux. Even after mounting a tmpfs and using it in neo4j i am still getting around 300tps only (((
I mounted a tmpfs in /etc/fstab
tmpfs /mnt/data1 tmpfs defaults,noatime,nosuid,nodev,noexec,mode=1777,size=2G 0 0
I'm using Jboss(WildFly 9) to host my Solr application on a Windows Server(2012), but if I start JBoss it's very slow and taking at least 15 mins to start. There is only one app (solr.war).
The same JBoss in my local with the same WAR takes less than a minute to start.
The JVM is set as MIN 4GB and Max 16GB on the servers.
What could be the reason it is very slow?
Try starting the sever without war and if it takes normal time in bootup then check the war code else check the jboss version and logs.
I have developed a Fusion Web Application on Oracle ADF.
Now i have deployed this Application on Weblogic Linux 64bit server and test this application using JMeter.
Initially i have test this application with 50 users then 100 and 500 users.
Problem is this, Server Memory and Swap space is not getting free.
It is getting increase even i run the 10 users test multiple times