On a workflow you can have 2 kinds of notifications:
You can configure user notifications to send email with different parameters. Inside the body of the notification you can customise the message thanks to the CDS variable templating with syntax like {{.cds.myvar}}
. You can also use HTML
to customise the message, then in order to let CDS interpret your message as an HTML
one you just need to wrap all your message inside html tag like this <html>MyContentHere</html>
.
You can configure for which node in your workflow CDS have to send a status on your repository service provider (Github, Bitbucket, …). You can configure if you want to have a comment on your pull-request when your workflow fails or you can just disable pull-request comment to only have status of your pipelines. By default you already have a default template for your pull-request comment but you can customize it with different kinds of templating. To have access about the node run
data and write some loops and conditions you can use the standard syntax as the go templating but with [[
]]
delimitters. You can also use the CDS interpolation engine with the same syntax you already know and use inside pipelines, for example: {{.cds.workflow}}
to get the name of the workflow.
For the go templating you have few variables you can use/iterate over.
.Stages
: an array of stages with .RunJobs
inside which are the array of runned jobs with their .Name
.RunJobs
: Inside a stage object which are the array of runned jobs
.Job.Action.Name
: The name of the runned job.Job.Status
: The status of runned job.Tests
: array of tests results
.Total
: total number of tests.TotalOK
: total number of OK tests.TotalKO
: total number of KO tests.TotalSkipped
: total number of skipped testsIf you need to know about other variable you can check WorkflowNodeRun
data structure here.
For example by default the template of pull-request comment is:
[[- if .Stages ]]
CDS Report [[.WorkflowNodeName]]#[[.Number]].[[.SubNumber]] [[ if eq .Status "Success" -]] ✔ [[ else ]][[ if eq .Status "Fail" -]] ✘ [[ else ]][[ if eq .Status "Stopped" -]] ■ [[ else ]]- [[ end ]] [[ end ]] [[ end ]]
[[- range $s := .Stages]]
[[- if $s.RunJobs ]]
* [[$s.Name]]
[[- range $j := $s.RunJobs]]
* [[$j.Job.Action.Name]] [[ if eq $j.Status "Success" -]] ✔ [[ else ]][[ if eq $j.Status "Fail" -]] ✘ [[ else ]][[ if eq $j.Status "Stopped" -]] ■ [[ else ]]- [[ end ]] [[ end ]] [[ end ]]
[[- end]]
[[- end]]
[[- end]]
[[- end]]
[[- if .Tests ]]
[[- if gt .Tests.TotalKO 0]]
Unit Tests Report
[[- range $ts := .Tests.TestSuites]]
* [[ $ts.Name ]]
[[range $tc := $ts.TestCases]]
[[- if or ($tc.Errors) ($tc.Failures) ]] * [[ $tc.Name ]] ✘ [[- end]]
[[end]]
[[- end]]
[[- end]]
[[- end]]
Which, for a pipeline with 1 stage and a job in failure, is displayed like this:
CDS Report build#11.0 ✘
* Stage 1
* my job ✘
And displayed on GitHub:
If you need to trigger some specific actions on the technical side, like for example use a microservice which listens to all events in your workflow (updates, launch, stop, etc.), you can add an event integration like, for example, Kafka and listen to the kafka topic to trigger some actions on your side. Events are more like sending notifications to machines instead of user notifications which are made for users. The see structure of sent events, you can look here and here.