Accessing Metrics and Building Reports in Cisco

advertisement
Tutorial
Accessing Metrics and Building Reports in Cisco
Metapod™—With Help from Graphite and Grafana
This document explains how to use Grafana to access the Graphite metrics provided
by Cisco Metapod, and shows you how to build custom monitoring dashboards based
on those metrics.
Cisco Metapod Metrics
Cisco Metapod is instrumented with many monitoring probes both for the health of the platform as well as to
monitor the performance of the bare metal and virtual instances. There are probes that monitor health alerts and
alarms that allow Cisco to proactively manage the cloud. Metapod uses Graphite to monitor statistics and
performance of the platform in different areas, which are displayed in the custom Horizon dashboard. Below are
some examples of metrics exposed through that dashboard. To navigate to the panel below, go to the hypervisor
or controller option under Admin, then click the caret on the left side of the name to pull out the metrics drawer for
viewing. This can be done one node at a time.
Hypervisor and Controller Node Bare Metal Stats
Figure 1 shows the metrics provided as part of the platform to monitor the performance of the bare metal hardware.
Hovering over the graphs shows the actual reading of that particular metric. Data is retained for up to a year and
historical views can show trends from a period of over a 1 hour, 1 day, 1 month, and 1 year timeframe.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 1 of 12
Hypervisor and Controller Bare Metal Live Stats
As part of the monitoring and metrics dashboard, “live stats” are exposed, which provide another view into the
bare-metal performance of the cloud. This feature provides a summary view of key metrics as well as drill-down
views of the “top,” “iostat,” and “netstat,” UNIX commands. The data is refreshed every five seconds.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 2 of 12
VM Stats
The next level of granularity is the VM-level status. This allows monitoring of an individual instance’s vCPU,
memory, and network performance. Below is an example of how these are represented in our dashboard.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 3 of 12
Storage Statistics
Storage statistics for the cloud begin with a high-level performance view of the entire Ceph storage pool. From
there the cloud administrator can drill down to the different pools, OSDs, and disk drive-level health charts.
Storage performance
The storage performance view gives users an understanding of how the overall Ceph storage pool is performing
from a throughput, IOPS, latency, and total storage utilization standpoint. Hovering over any metric gives additional
details about it and provides a view over a period of time for historical purposes. This view spans the entire storage
pool.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 4 of 12
Pool Statistics
This option allows cloud administrators to double click into a pool to analyze performance. The example in the
figure below shows two pools: one for instances and one for the Glance image store.
Ceph OSD Health
The Ceph health view allows the administrator to see if there are any Ceph object store daemons (OSDs) or disks
that are down or out of the Ceph pool. Cisco support will alert cloud administrators if failed drives are detected or if
OSDs go offline. However, this view allows the administrator to verify operational health.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 5 of 12
Accessing the Graphite Dashboard
Graphite is an open source tool for monitoring and gathering metrics. To access the Graphite dashboard, visit the
URL for the Metapod cloud and add /graphite to the end of the URL. The screen below will appear.
http://customer-example-dashboard.client.metacloud.net/graphite
From there, users can see the monitoring already set up within the platform and examine each item. Users can
also create their own charts under the User Charts option as seen on the snapshot below. Many users choose to
use Grafana to view the different metrics collected through a tree. The next section will discuss how to install
Grafana on a Linux VM or server, point it to a Graphite API end point, and collect data to build custom graphs.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 6 of 12
Installing Grafana
Grafana is an open source software project that has direct support for Graphite APIs as well as other data sources
like Elasticsearch and a few others. Since it’s an easy way to enhance the data shown in the Graphite GUI, many
customers will use Grafana to gather data from the Graphite instance and create custom graphs and reports. The
section below walks through installing Grafana on Ubuntu and configuring it to talk to the Metapod Graphite APIs.
The first requirement is a Linux environment upon which to install the Grafana software. In this example an Ubuntu
server is used to run Grafana. Below are the instructions showing how to install the package for Grafana:
Once the server is built and the software installed, ensure port 3000 is open for http access then connect to it with
a browser at its IP address or name. The login screen to access Grafana is shown below. Once logged in, the
platform can be customized and directed to the API endpoints. The default login/password is admin/admin. After
the user is logged in, the first step is to add a data source to build custom graphs.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 7 of 12
Point Grafana to the Metapod Graphite URL using the data sources option as seen in the screen shot below.
Select the Graphite options form the menu and enter the URL of the dashboard including the /graphite addition.
Grafana supports other backends as well as Graphite. Select the data source and add it.
The next step is to start building reports for various dashboards with the different metrics desired by the end user.
The example below demonstrates how to build a two-panel report. The first panel will look at CPU load on three
controller nodes. The second will look at memory on all three nodes. This leverages built-in Metapod agents to pull
data into Grafana.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 8 of 12
To build the report, first create a new dashboard and the two panels described above. Simply add a new
dashboard via the GUI and give it a name (e.g. Trial2 CPU). Next add the panel called CPU Utilization and make it
a graphical panel as shown in the screenshot below. This will provide the ability to point to the Graphite instance
monitoring Metapod and extract data from it.
Once the panel is added it will take up the whole row. It can be edited and made smaller using the “-“ on the Edit
menu for the panel. A customized panel from data and x-axis and y-axis perspective can be created and even
dragged around on the screen.
After adding the graphical panel, give it a name and set up the metrics to monitor. The example below shows the
metrics Carbon:*:*:cpuUsage. This was able to graph all the CPU usage for all three controllers on one graph.
CPU Utilization as a Y-axis label is also shown in this example.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 9 of 12
Save this panel and make it smaller to create a memory panel next.
Use the “Add Panel” option to add a panel as done previously but call it “MCP Memory” to look at memory usage.
This last example shows what you can do with a little creativity when working with Graphite and Grafana. By
adding more panels to the top row, we can now see more of the data at once, plus we’ve added a second row,
comprised of boxes containing critical single metrics. Furthermore, we’ve set thresholds for each of these metrics
so the colors will change as the metric changes values—therefore making it easier to evaluate system performance
at a glance.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 10 of 12
References
•
Cisco site: http://www.cisco.com/c/en/us/products/cloud-systems-management/openstack-privatecloud/index.html
•
Cisco Metapod Support site: https://support.metacloud.com/forums
•
Release Notes: https://metacloud.zendesk.com/entries/95023728-Cisco-OpenStack-June-2015-Update
•
Graphite Docs site: https://graphite.readthedocs.org/en/latest/
•
Grafana Docs site: http://docs.grafana.org/
For More Information
Visit our website to read more about Cisco Metapod features and benefits.
To access more technical tutorials about this product, visit our Community page.
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 11 of 12
Printed in USA
© 2015 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 12 of 12
Download