Hello everyone!

I have used the community knowledge in numerous times. And this is my first post on this community, since, for the first time, that I can not find an answer.

We have deployed in VMWare vSphere vCenter 6 some CentOS machines, but we're having problems with only one. The machine has 8 threads (4 sockets with 2 cores each), with 16GB of ram.

On this machine we deployed Tomcat7, Oracle JRE1.8, MySQL, etc., to deploy an application of a database POC.

After some working time, and this is happens randomly, the memory usage increases to almost all the ram available, the server starts being very slow responsive, and after a while, it starts show the warnings on the console and / var / log / messages "Kernel: NMI watchdog: BUG: soft lockup"

One of the odd things is that, at TOP it shows almost 11GB of RAM used, 5GB in cache and only 210MB free, with no swap usage. But, there's only one pair of process using ram, and adding up everything that's showed at ps, I got only 5GB, and I do not know the rest of the ram is going to. Even stopping tomcat and mysql, the numbers does not change and the memory usage keeps the same.

I triyed to run this on CentOS7 1511 and 1611, and now I'm back on 1511, but, with kernel 4.11.3-1.el7.elrepo.x86_64 installed through yum.

Since I couldn't solve this by now, I've downloaded the Turnkey Tomcat appliance, and it is running the same POC application smoothly and way faster.

I'll appreciate any help. Thanks in advance.

If you are getting "CPU soft lockup" on VMWare, it means that the VMWare hypervisor is not dispatching your VM.

The problem is not with the resource picture as seen by your VM, but rather the truth as seen by VMWare.

1 members found this post helpful.

Hello! Thanks for your answer!

Well, I have to ask you what do you mean with that, and how can I go after it, because these vm's are hosted in a recently bought Dell blade, and there's not a single performance alarm on the vcenter. In fact, checking the performance charts, the average CPU usage on the blade is bellow 30% , and the use on the VM itself when the problem shows, is usually very low too.

Thanks for any information.