I am using the release pipeline classic editor and have a requirement of passing an output variable generated in a task to multiple jobs in the same stage or to outside stages. Currently, this output variable is available only inside the same job and I have to write the same task in multiple jobs and stages and I feel it is a redundancy. Is there any way to implement it?.
In the Classic editor, I am afraid that output variables in a different job is not feasible. Please refer to this document .
As a workaround , you can use variables across jobs and stages via variable groups.
First define the variable in the variable group, then update the variable group through rest api or azure cli, and replace the defined variable with the value of the variable generated by the task.
PUT https://dev.azure.com/{organization}/{project}/_apis/distributedtask/variablegroups/{groupId}?api-version=5.1-preview.1
Here is a case about update variable group with powershell script.
Another workaround : You can share values across all of the stages by using release pipeline variables. The solution is updating the Release Definition for the Release Pipeline variable in the Stage where the variable is set.
Define a variable in the release definition Variable.
Use REST API Definitions - Update to update the value of the
release definition variable in the agent job.
Use the updated value of the release definition variable in the next
agent job.
The details info about using REST API to update the value of the release definition variable, you can follow this ticket.
For detailed steps and guide, please refer to this blog .
Related
I have a reasonably complex release pipeline in Azure DevOps that releases a number of Azure apps, a database etc.
Each step is genericised using a library variable for the environment. For example:
But library variables are linked to a release or a selection of stages.
Currently I have to clone the entire pipeline and link a new library variable group in the clone to publish a different environment, but this is heavy on unwanted duplication and maintenance.
How can I run the same release pipeline with different library variables?
If I could do this, it would be possible to have a release for a given branch, for example, but I cannot see a way to do it.
At of this time, it is not supported to select which variable groups to use when you create a release.
If you only have one or several variables, I think you can use pipeline variables instead of variable groups, so that you can update them at release time. Here are the detailed steps:
Go to your pipeline editing page and select "Variables" tab. Click "Add" to add a variable. Then check the option "Settable at release time".
Try re-create your release. You will find the variables defined in #1 and you can edit them before create the release.
If you have many variables, I suggest you try to change the structure of your pipeline to make it more suitable for deployment to multiple environments. As Daniel said, you can use stages for each environment, and then use the variable group in stages scope.
I was looking for documentation on when Azure DevOps 'freezes' the variable values for pipelines. This is mainly important when updating variables, either pipeline variables or within variable groups. As far as I know, DevOps creates a copy of the variable at some point so that if you redeploy a pipeline or change a variable in the meantime, you do not get the updated value for existing releases. But I'm not sure at which point in time that variable values are frozen, nor whether this behavior is equal for variables defined in the pipeline (for the entire release or for a specific stage) and variable groups.
If anyone could either point out the relevant documentation (which I was not able to find myself) or give me some insights into how this works, that would be great!
I have 2 build pipelines in my azure devops project, one for building source code and the other one is for
making the setup.
I want the build number generated by the first pipeline that compiles code to be passed to the next pipeline which creates the setup file because i want the setup file to take the same version, so I added a variable group with a variable called sharedBuildCounter.
But when I set sharedBuildCounter the build number in the first pipeline using logging command like this(used inside PowerShell task):
Write-Host "##vso[task.setvariable variable=variable_name;]new_value"
The variable indeed takes the new value and I am able to output the new value using another PowerShell task with one line:
Write-Host $(SharedBuildCounter)
And when I run the next pipeline that builds the setup, I find that sharedBuildCounter is being re-set to the default empty value.
Notice: I found threads that suggests using API rest calls to change variable values, but it don't seem to include a specific pipeline name in case of using pipeline variables(not variable groups).
Variable groups will help to share static values across builds and releases pipeline.
What you need is a way to pass variables from one pipeline to another. I'm afraid to say the is no official way to do this.
As a workaround you could update the value of your variables inside your variable group. There are multiple ways to handle this, Rest API, powershell, 3rd-party extension. Detail ways please refer answers in this question: How to Increase/Update Variable Group value using Azure Devops Build Definition?
If you want to get the value of variable in the pipeline. Since you have used logging command to update that variable.
You need to use Rest API to get that particular build log to fetch related info.
You can use Azure Artifacts to pass information between pipelines. In one pipeline, you write the values to a file and publish the file to an artifact. In the other pipeline, you download the artifact and read the file.
There may be other ways to do it. Azure DevOps allows for free and infinite use of Azure Artifacts in this fashion.
See How to get variable values from pipeline resources in azure pipelines.
I have two stages in my pipeline. The first one is the trigger for the second one. I want the parameter from first stage be input/accessible in the second stage.
Is this feasible?
EDIT
The case is when the parameter value is set on stage 1 as of result of ARM or script output, then this value is not visible on next stages.
Thanks
It is not possible this way to share variable from stage1 to stage2 in case when you are changing variable value in stage1, to achieve this you need to persist this value in some storage like for eg. keyvault, azure function or vsts api that you will change value for . The stage1 and stage2 can be running on different agent. What can be done is to edit variable value for stage2, in stage1 add task that is persisting this value using (vsts/tfs/azure devops) api and persisting this on release definition. Api for update release definition https://learn.microsoft.com/en-us/rest/api/vsts/release/definitions/update?view=vsts-rest-4.1
Would anyone know if there is any magic way to define an env variable that could be accessed from all stages of the DevOps build pipeline?
So far, it seems the only thing you can do is to define env variables shared by all jobs inside the same stage.
I understand that each stage run in complete isolation under the cover, but still I was wondering if any trick could be done to define a common env variable for all stages?
Thanks
Environment variables are per stage. There is currently no way to have the same environment variable across that entire pipeline. You would need to set it on each stage individually.
Environment variables are per stage, correct, so if you want one to exist across all stages in a pipeline, you'll need to add it to each stage.
For a list of built-in environment variables and other information, take a look at this documentation page: https://console.ng.bluemix.net/docs/services/DeliveryPipeline/deploy_var.html