Open yaroslav-nakonechnikov opened 2 weeks ago
To increase the number of TCP connections within a container, you may need to adjust several kernel parameters and container resource limits. Here are steps to help you achieve this:
somaxconn
in the Host Cluster NodeThe somaxconn
parameter determines the maximum number of connections that can be queued for acceptance. You already have somaxconn
set to 4096, but you may want to increase this further on the host:
sudo sysctl -w net.core.somaxconn=65535
You can make this change permanent by adding it to /etc/sysctl.conf
:
echo "net.core.somaxconn=65535" | sudo tee -a /etc/sysctl.conf
sudo sysctl -p
Depending on the container runtime, the container may inherit the host's somaxconn
setting. However, it can also have its own limits:
For Splunk Pod You can set sysctl
parameters at the Pod level using the securityContext
in your Pod manifest:
apiVersion: v1
kind: Pod
metadata:
name: splunk
spec:
securityContext:
sysctls:
- name: net.core.somaxconn
value: "65535"
containers:
- name: your-container-name
image: your-image
Since you can't pass the securityContext directly to the Splunk pod through the Splunk Operator custom resource at the moment, you might consider these alternative approaches:
Alternative Approaches Apply Settings on the Host: If feasible, you could set the necessary kernel parameters at the host level (if you control the host machines), which the containers inherit. This wouldn't provide per-container granularity but would solve the immediate need to increase connection limits.
we will work on enhancing Custom Resource to take secruitycontext , thank you
so, there is no setting for it, and need to create pre-task playbook to make it work, right?
ps. making setting on host is not applied to pods in kubernetes. it should be allowed explicitly:
hello, we see this:
this looks a bit strange and we would like to understand how it would be possible to increase the limits?
this is splunk container on kubernetes engine.