[caasp-beta] How-To: Container Monitoring and Performance Analysis

[Alejandro Bonilla]

Hi Cameron,

That is a nice question for our product managers. I believe that may be our intention in an upcoming release.

Certainly, we’d want our customers to use containers that were built by a trusted vendor or with Dockerfiles/container images provided by a trusted source.

I’d modify the traditional saying to, “Rome wasn’t built by Beta2”

Thanks,
Al
 

> On Apr 12, 2017, at 3:21 PM, Cameron Seader <CSeader at suse.com> wrote:
> 
> Would be nice to have these things in our own container registry. :-)
> 
> 
> 
> Sent from my Verizon, Samsung Galaxy smartphone
> 
> 
> -------- Original message --------
> From: Alejandro Bonilla <abonilla at suse.com>
> Date: 4/12/17 11:58 AM (GMT-06:00)
> To: caasp-beta at lists.suse.com
> Subject: [caasp-beta] How-To: Container Monitoring and Performance Analysis
> 
> Hi Beta Testers -
> 
> (Before you proceed, please refer to my previous e-mail, "A few How-To guides coming your way”)
> 
> —
> 
> We need to have some way to understand how/what our cluster is doing around resource consumption (CPU, Mem, Disk, Network, etc). One option is Heapster.
> 
> Heapster enables Container Cluster Monitoring and Performance Analysis, using InfluxDB and Grafana.
> 
> These steps help deploy Grafana+InfluxDB+Heapster. The Grafana Dashboard helps visualize the data collected into InfluxDB.
> [kube-dns is required for proper DNS resolution, an a LoadBalancer or my previous e-mail on using an LB should help to facilitate access to the dashboard]
> 
> 1. Git Clone the repo that contains the config files
> 
>         git clone https://github.com/kubernetes/heapster.git
> 
> 2. Open/modify the heapster deployment so that it uses the FQDN vs short name, as our certificates are only set for full names (Changing in an upcoming release).
> 
>         cd heapster
>         vi deploy/kube-config/influxdb/heapster-deployment.yaml
> 
> 3. Modify line 20 to look like
> 
>         - --source=kubernetes:https://kubernetes.default.svc.cluster.local
> 
> 4. Deploy All the Services and Deployments
> 
>         kubectl create -f deploy/kube-config/influxdb/
> 
> 5. Deploy a configmap to expose ‘monitoring-grafana’ service via the LoadBalancer. [This depends on kube-loadbalancer-rc from a previous e-mail]
> 
>         Save the attached file
>         kubectl create -f grafana-configmap-dashboard.yaml
> 
> 6. Get the IP of the LoadBalancer for service ‘monitoring-grafana’ defined as 'grafana-configmap-dashboard-svc'
> 
>         kubectl get configmap grafana-configmap-dashboard-svc -o yaml
> 
> 7. Access the Grafana Dashboard via the provided bind-ip:
> 
> If ever asked for a username/password, it should be root:root
> 
> Please let me know if anything went wrong.. or right.
> 
> Thanks,
> 
> Alejandro Bonilla
> SUSE
> Technical Strategist – North America
> abonilla at suse.com
> (484)426-7722