Hi all, I have set up distributed jmeter client-server setup to run performance tests. However, I am reaching a limit of *1849 req/s with 25K threads/ users* (RampUp = 5s) when not using client-server setup. (for a simple GET request)
I would like to use single machine to generate maximum load and throughput before moving to master-slave configuration. I configured the c4.xlarge EC2 instances by setting following values: /proc/sys/net/core/somaxconn change to 50000 /etc/sysctl.conf add the following lines: # allow more than the default 128 max network connections net.core.somaxconn=50000 # allow more network connections net.netfilter.nf_conntrack_max = 256000 # allow more open files fs.file-max = 100000 increase file descriptor limits for each user: /etc/security/limits.conf # add the following two lines * soft nofile 30000 * hard nofile 100000 What is the maximum load that I can generate in jmeter using c4.xlarge instance? Do I have to modify any other setting on jmeter or EC2? java heap size? Please let me know if you need more information. Thanks! Meena
