My servers don't have access to external resources (gems, etc), so I manually run
bundle package
then upload the code and run
bundle install --local
How can I get the same behaviour out of Capistrano? Using the Bundler recipe, tries to install the bundles on the server. Whereas I'd like to;
On the build machine
Checkout from SVN
run bundle package
zip and upload the artifact
On the application server
Expand the Zip artifact
run bundle install --local
start the server
I tried
after("deploy:update_code") do
system("cd #{copy_cache} && bundle package")
end
But that runs the checkout and upload, there didn't seem to be a place to attach after the checkout, but before the zip.
It seems the best way to do this is to, run bundle package on the development workstation, and then commit the bundled gems to source control, then run;
bundle install --deployment
Related
I'm trying to build a theme app extension using CLI 3.0, I managed to create the extension on my app but I have 2 basic issues:
Whenever I add a js file to the assets folder or a liquid file to the blocks folder the build fails, even if the file is empty (build passes if I remove the file) - this is the error I get when running npm run build
Command failed with exit code 1: bundle exec theme-check /Users/omerdotan/projects/shopify-omer-test-app-2/extensions/omer-theme-extension-test-2 -C :theme_app_extension
The app doesn't appear on the theme - I attached the app to the store and enabled the developer preview for the theme extension,but it show no apps available (for both App embeds on theme settings and blocks)
I was running into similar issues.
My solution:
Install rbenv https://github.com/rbenv/rbenv, set required ~/.zshrc file using nano ~/.zshrc
Use rbenv to install latest Ruby version (3.1.2)
Target rbenv version to latest version by rbenv global 3.1.2
Reopen terminal and gem install bundler
Reopen shopify terminal and npm run deploy
I'm working on creating a local repository that will contain all packages I use in my project, so I can have those packages installed on a machine that does not have access to the internet. I think of the repository that I could clone on the machine and run yarn install to have all the packages available in the project from the local repository. How can I do that? Similar question was asked here Using npm how can I download a package as a zip with all of its dependencies included in the package
There's not enough information in your question to fully understand your situation, but if you commit your node_modules directory to the repository, the modules will be there without the user having to run npm or yarn to install them. This assumes the user will run code from the repo workspace and that there aren't any modules that require a compilation step or other build step that may be platform-specific. But if they're all plain ol' JavaScript modules, you should be fine.
If you want to have all the modules as a separate repo rather than checking in node_modules, I can offhand think of two ways this might work.
Have the packages repo be a check-in of a fully installed node_modules directory. Then make that repo a Git submodule of the main repo that gets cloned as node_modules in the main repo.
Use npm pack to create .tgz files for each package you need. Store those files in the packages repo. Clone that repo into a known path on your target machine. Have the main repo install via path names. For example, if you run npm install /var/packages/foo-1.0.0.tgz, it will add a line to your package.json that might look something like this: "foo": "file:../../../var/packages/foo-1.0.0.tgz". In that case, npm install will install from that path rather than over the network.
I have a React Native app hosted on Microsoft App center. The builds (both iOS and android) are failing because of yarn 1.19 (error Incorrect integrity when fetching from the cache)
I tried to remove the package-lock.json but it didn't help.
I would like to downgrade yarn or execute the cache clean command but don't know where to execute it.
I have installed the appcenter cli version 2.2.1 and successfully connected to it.
Where could I execute yarn cache clean for example?
I've read I could also create a script but I have no idea where to place it and how it should look like. Should it be both in the ios and android directory? Or in the root? Thank you
I actually noticed that I had a yarn.lock back from the initial commit, even though we are using npm in the team.
In the build logs, there is this line:
/bin/sh -c if [ -f yarn.lock ]; then { yarn install && yarn list --depth=0; } else npm install; fi
So I deleted the yarn.lock file and now it builds successfully using npm!
Not sure whether you build your React app with Azure Devops pipeline. If yes, you can use Command line task to achieve the things you want to do.
If the agent you used is hosted agent during the build, since each build will use a completely new VM, you need to install the AppCenter cli once per build.
Use follow command to install the AppCenter cli:
sudo npm install -g appcenter-cli
Then logging in it:
appcenter login --token {Token}
Here, you need first generate the token with this doc described: Go and login to https://appcenter.ms/ -> Click Self Avatar -> Choose Account Settings -> Click on API Tokens -> Click New API token then select the corresponding the scope for this token.
Copy and use it in this pipeline task. Note: Recommend you store
this token with secret variable for security.
Now, you can execute the clean command: yarn cache clean.
Where to place it and how it should look like?
According to your description, you need place this command line task into the first step, then it could clean the Yarn cache firstly.
Also, because of the image configuration that the hosted agent is using, its installed node.js version is 6.7.0, this does not match the runtime environment for AppCenter cli. You need also run Node.js tool installer task to install node.js v10.0.0.
All of them should look like this:
Should it be both in the ios and android directory?
As I mentioned previously, for Hosted agent, each build will use a completely new VM. So yes, you must execute these two steps firstly in every build.
If what you used is your private agent, since you have installed the AppCenter cli locally, the agent will automatically call the local configuration when running the command line task. At this time, you just need to skip the install command in the command task:
We fixed it by adding a "yarn cache clean" in appcenter-post-clone.sh, you can add this shell script in root of project.
See this docs for details.
I've got a react-native app that builds on AppCenter. I have a Podfile in my /ios directory, and locally, I can run pod install. The AppCenter documentation says that "App Center scans the selected branch and if it finds a Podfile, it will automatically do a pod install step at the beginning of every build. This will ensure that all dependencies are installed.", but as far as I can tell, it's not running pod install for my build and the build fails for that reason.
Tried adding pod install to a pre-build script, in which case the build succeeds.
Make sure the Cocoapods version running on the build agent matches the one on your podfile.lock.
You can add a post clone script like this on your ios/ folder:
appcenter-post-clone.sh This will automatically install the matching cocoapods version.
Push that file to your repo, then make sure you open the build configuration and hit Save and Build.
It appears that in addition to having a Podfile, you must NOT have a ios/Pods folder present. In my case, I also had to explicitly add that folder to my .gitignore
I need to create node_modules in the root app when deploy new version using Octopus
How can I create a run script using a new octopus step? I try:
npm install
But I get error when execute the script step
'The remote script failed with exit code 1'
Any ideas?
Your build system should be handling that for you. The node_modules directory should then be included in your application package, but ideally, this should also be bundled to reduce the number of files that your application is dependant on.
If you do need to run this as part of a deployment, then you would need NodeJS and npm installed on the server that is executing the deployment step (either the Octopus server itself or a worker instance if using workers).
For more information check out this blog post