Open Boolos opened 7 years ago
how often are you seeing this, and is this a new thing or has this been an intermittent issue for awhile? we don't currently have a way for you to add a retry yourself, but we can investigate adding that as a new feature.
@Boolos if you need shippable retry until we implement the feature, you will need to remove the hub integration and login to docker using commands in YML. Your AWS credentials can be setup in a key-value pair integration (which will set the right environment variables) for the AWS CLI to use or as secure env variables. The AWS CLI will also have to be installed using straightforward pip/apt-get commands.
If you need an example yml, please let me know.
@trriplejay it's been a relatively recent issue. it may have to do with some of the Amazon AWS instability lately? and it's not a huge deal since we can manually restart the build, and it usually runs to completion. just would be nice not to have the false fails, hehe.
@ambarish2012 i'll do that as a work around if it becomes a frequent occurrence. thx for the suggestion!
@Boolos thanks. We are adding this feature request to our backlog and we agree that our platform should implicitly retry connections to third party service. I will let you know when this feature goes live in production.
awesome! thx @ambarish2012 👍
Occasional build failure due to transient network issues
CI tasks within the
shippable.yml
can be prepended withshippable_retry
in order to mitigate transient network issues. However, my CI builds occasionally fail when trying to login to a hub integration (e.g. Docker Hub). I was wondering if there's anyway to specify the equivalent ofshippable_retry
but for integrations, i.e. this section of the .yml: