procwatch

procwatch

Watch processes and report their usage consumption (CPU, memory).

License

Apache v2

Dependencies

• gopsutil
• kubernetes APIs

Installation: kubernetes/kubevirt cluster

This project can be deployed in a kubevirt cluster to report metrics about processes running inside PODs. You may use this to monitor the resource consumption of these infrastructure processes for VM-based workloads. We assume that the cluster is running prometheus-operator to manage the monitoring using prometheus, and kubevirt >= 0.9.1, which includes itself better integration with prometheus operator.

1. First deploy a new service to plug in the configuration KubeVirt uses to interact with prometheus-operator:

kubectl create -f procwatch/cluster/kube-service-vmi.yaml

1. Now deploy the tool itself into the cluster: Set PLATFORM to either "k8s" or "ocp" and

kubectl create -f procwatch/cluster/collectd-config-map.yaml
kubectl create -f procwatch/cluster/collectd-node-agent-$PLATFORM.yaml

1. procwatch installs a new deployment in the kube-system namespace. VM pods usually run in the default namespace. This may make the prometheus server unable to scrape the metrics endpoint. procwatch added. To fix this, deploy your prometheus server in your cluster like this:

apiVersion: monitoring.coreos.com/v1
kind: Prometheus
metadata:
  name: prometheus
spec:
  serviceAccountName: prometheus
  serviceMonitorNamespaceSelector:
    matchLabels:
      prometheus.kubevirt.io: ""
  serviceMonitorSelector:
    matchLabels:
      prometheus.kubevirt.io: ""
  resources:
    requests:
      memory: 400Mi

Note the usage of serviceMonitorNamespaceSelector. See here for more details

1. Now, you may need to add labels to the namespaces, like kube-system. Here's an example of how it could look like:

apiVersion: v1
kind: Namespace
metadata:
  ...
  creationTimestamp: 2018-09-21T13:53:16Z
  labels:
    prometheus.kubevirt.io: ""
...

1. The last step: now you need to deploy a Service Monitor to let prometheus-operator pickup and add rules for this endpoint:

kubectl create -f procwatch/cluster/kube-service-monitor-vmi.yaml

Please check the next sections for Caveats.

Notes about integration with kubernetes/kubevirt

Please be aware that in order to resolve the PIDs to meaningful VM domain names, procwatch needs to access the CRI socket on the host. This is equivalent of exposing the docker socket inside the container. This may or may not be an issue on your cluster setup.

If you disable the CRI socket access, procwatch will just report the PIDs of the monitored processes.

Installation: bare metal

1. Make sure you have the golang toolset installed on your box. For example, on Fedora:

   # dnf install golang-bin
   

If this is your first golang application, make sure you have the GOPATH set:

$ export GOPATH="$HOME/go"

You may want to make this setting persistent

1. (TODO): ensure the vendored dependencies

2. checkout the sources, and transparently build the tool

$ go get github.com/fromanirh/procwatch

1. copy the tool on your filesystem:

$ sudo cp $GOPATH/bin/procwatch /usr/local/libexec

1. fix the SELinux configuration:

# semanage fcontext -a -t collectd_exec_t /usr/local/libexec/procwatch
# restorecon -v /usr/local/libexec/procwatch

1. copy the recommended configurations:

# mkdir /etc/procwatch.d
$ sudo cp $GOPATH/src/github.com/fromanirh/procwatch/conf/procwatch/*.json /etc/procwatch.d/

1. copy the collectd configlets:

$ sudo cp $GOPATH/src/github.com/fromanirh/procwatch/conf/collectd/procwatch*.conf /etc/collectd.d/

1. restart collectd

# systemctl restart collectd

1. Done!

   # collectdctl listval | grep exec
   kenji.rokugan.lan/exec-vdsmd-4615/cpu-perc
   kenji.rokugan.lan/exec-vdsmd-4615/cpu-system
   kenji.rokugan.lan/exec-vdsmd-4615/cpu-user
   kenji.rokugan.lan/exec-vdsmd-4615/memory-resident
   kenji.rokugan.lan/exec-vdsmd-4615/memory-virtual
   kenji.rokugan.lan/exec-vdsmd-4615/percent-cpu