Documentation / Metrics
Sitespeed.io collects a lot of metrics which are filtered before they are sent to Graphite/InfluxDB. You can remove filters and/or add your own filters. Some sensible defaults have been set for you, if you have suggestions to change them create an issue at Github.
The metrics are separated into two groups:
The pageSummary encapsulates the metrics for a single page. For example if we test a single page 10 times, it will have the min/median/max here for values that change.
The summary holds information per group, or specifically per domain. If you test ten different pages on a site (same domain), the summary of those metrics will end up here.
You can list the metrics that are configured by --metrics.filterList. The list is dependent on which plugins you are loading, so you will need to do an actual run to generate the list. The list is stored in the data folder in a file named configuredMetrics.txt.
docker run --shm-size=1g --rm -v "$(pwd)":/sitespeed.io sitespeedio/sitespeed.io https://www.sitespeed.io --metrics.filterList
The file will look something like this:
browsertime.pageSummary.statistics.timings.timings
browsertime.pageSummary.statistics.timings.rumSpeedIndex
browsertime.pageSummary.statistics.timings.fullyLoaded
browsertime.pageSummary.statistics.timings.firstPaint
browsertime.pageSummary.statistics.timings.userTimings
browsertime.pageSummary.statistics.visualMetrics.SpeedIndex
browsertime.pageSummary.statistics.visualMetrics.FirstVisualChange
browsertime.pageSummary.statistics.visualMetrics.LastVisualChange
browsertime.pageSummary.statistics.custom.*
...
You can also list all possible metrics that you can send. You can do that by using --metrics.list. It will generate a text file named metrics.txt in the data folder.
docker run --shm-size=1g --rm -v "$(pwd)":/sitespeed.io sitespeedio/sitespeed.io https://www.sitespeed.io --metrics.list
You can add/change/remove filters with --metrics.filter.
If you want to add metrics, start by looking at the generated metrics file, so you can see what you would send.
Say you want to send all performance advice to Graphite for a page. Checking the file you should see them. It will look something like this:
...
coach.pageSummary.advice.performance.adviceList.avoidScalingImages.score
coach.pageSummary.advice.performance.adviceList.avoidScalingImages.weight
coach.pageSummary.advice.performance.adviceList.cssPrint.score
coach.pageSummary.advice.performance.adviceList.cssPrint.weight
coach.pageSummary.advice.performance.adviceList.fastRender.score
coach.pageSummary.advice.performance.adviceList.fastRender.weight
coach.pageSummary.advice.performance.adviceList.inlineCss.score
coach.pageSummary.advice.performance.adviceList.inlineCss.weight
coach.pageSummary.advice.performance.adviceList.jquery.score
coach.pageSummary.advice.performance.adviceList.jquery.weight
coach.pageSummary.advice.performance.adviceList.spof.score
coach.pageSummary.advice.performance.adviceList.spof.weight
coach.pageSummary.advice.performance.adviceList.thirdPartyAsyncJs.score
coach.pageSummary.advice.performance.adviceList.thirdPartyAsyncJs.weight
...
The score is … yes the score and the weight is how important it is. You probably only need the score, so setting a filter like this coach.pageSummary.advice.performance.adviceList.*.score will send them all (setting a wildcard for the name).
docker run --shm-size=1g --rm -v "$(pwd)":/sitespeed.io sitespeedio/sitespeed.io https://www.sitespeed.io --metrics.filter coach.pageSummary.advice.performance.adviceList.*.score -n 1
The best way to test and verify on your local, is to checkout the sitespeed.io project and then start a TCP server that logs everything:
tools/tcp-server.js
And you will see something like this:
$ Server listening on :::52860
It will output the port, so you can then use it when you run sitespeed.io:
docker run --net host --shm-size=1g --rm -v "$(pwd)":/sitespeed.io sitespeedio/sitespeed.io --metrics.list https://www.sitespeed.io -n 1 --metrics.filter coach.pageSummary.advice.performance.adviceList.*.score --graphite.host 127.0.0.1 --graphite.port 52860
The the previous example it will log all metrics you send to Graphite to the console.
By default the total score for performance, accessibility and best practice is configured to send to Graphite. Previously we looked at sending all the score for the performance advice. If you want to send all the scores for all advice, you can do that easily, by adding all three categories in the CLI:
--metrics.filter coach.pageSummary.advice.performance.adviceList.*.score coach.pageSummary.advice.bestpractice.adviceList.*.score coach.pageSummary.advice.accessibility.adviceList.*.score
If you have a lot of different metrics that you want to send to Graphite the command line will be overloaded. We have a solution to the problem: Use the JSON configuration option --config (read more here).
If you want to send the three coach metrics, you can add them as a config file like this and pass the filename to the --config parameter:
{
"metrics": {
"filter": [
"coach.pageSummary.advice.performance.adviceList.*.score",
"coach.pageSummary.advice.bestpractice.adviceList.*.score",
"coach.pageSummary.advice.accessibility.adviceList.*.score"
]
}
}
Sitespeed.io does not currently have support removal of a single metric, but you can remove all configured metrics with the parameter value *-. Here is an example sending only the coach.pageSummary.advice.performance.adviceList.*.score metrics.
docker run --net host --shm-size=1g --rm -v "$(pwd)":/sitespeed.io sitespeedio/sitespeed.io --metrics.list https://www.sitespeed.io -n 1 --metrics.filter *- coach.pageSummary.advice.performance.adviceList.*.score --graphite.host 127.0.0.1 --graphite.port 52860