This repo collects code, instructions and results for scalability tests on the Rancher product family.
See the docs directory for a list of tests and their usage specifics.
If you get this error from terraform apply
:
Error: Failed Cluster Start: Failed to add one or more agents: Node k3d-... failed to get ready: error waiting for log line `successfully registered node` from node 'k3d-moio-upstream-agent-0': stopped returning log lines: node k3d-... is running=true in status=restarting
And docker logs
on the node container end with:
kubelet.go:1361] "Failed to start cAdvisor" err="inotify_init: too many open files"
Then you might need to increase inotify's maximum open file counts via:
echo 256 > /proc/sys/fs/inotify/max_user_instances
echo "fs.inotify.max_user_instances = 256" > /etc/sysctl.d/99-inotify-mui.conf
If you get this error from terraform apply
:
│ Error: Kubernetes cluster unreachable: Get "https://upstream.local.gd:6443/version": dial tcp 127.0.0.1:6443: connect: connection refused
SSH tunnels might be broken. Reopen them via:
./config/open-tunnels-to-upstream-*.sh
In case Terraform returns an error with little context about what happened, use the following to get more complete debugging output:
export TF_LOG=debug
Use the following command to point to a remote Docker host:
export DOCKER_HOST=tcp://remotehost:remoteport
Note that the host has to have TCP socket enabled, in addition or replacing the default Unix socket.
Eg. on SUSE OSs edit the /etc/sysconfig/docker
file as root and add or edit the following line:
DOCKER_OPTS="-H unix:///var/run/docker.sock -H tcp://127.0.0.1:2375"