Your data. Anywhere you go.

New Relic for iOS or Android


Download on the App Store    Android App on Google play


New Relic Insights App for iOS


Download on the App Store


Learn more

Close icon

Uptime v Downtime

sla
sli
uptime

#1

Internally are using envoy to route traffic between different services.
simplistically Envoy knows what services run on what node.
We are using NR insight to calculate error rate and response time SLIs (ratios of bad or slow requests to all requests against a threshold).
One issue we have is when all service nodes are warming up they are not in service so we do not want to include them. Likewise if all service nodes are down we want the SLIs to show we are failed.
A simple query we are using is like…
SELECT filter(count(), WHERE duration >= .300)/count() * 100 AS PERCENTAGE FROM Transaction where appName = ‘XXX’ SINCE 24 hours ago
How do we figure in downtime?
We do not want to use Synthetics.


#2

Hi @feidhlim.oneill,

In the Insights product, the data provided there is entirely event data and metric data driven. These data are available for querying using NRQL when a language agent has successfully collected and sent them from your web node.

With that being said, if your host is currently in downtime (powered off or otherwise), the language agent process on the host is no longer collecting and actively sending data to New Relic. This means that no metric data or event data is being seen by the collectors and is not being recorded as having happened on your host, this also means that querying for these data inside of Insights would result in a null value.

Measuring the “heartbeat” or uptime of a host is best done from outside of said host. This is why we suggest using a Synthetics ping to determine uptime, which can then be queried for within Insights and added to a dashboard. Aside from a Synthetics ping check, I don’t believe uptime can be calculated using our APM and Infra agents, because both require running processes on the host for data to be recorded.

Let me know if you have any questions about this.

Thanks!