We believe that the best way to build software is to do it in close collaboration with the people who use it. We invite you to submit your ideas using the form below. Please be sure to include the problem for which you are solving and the benefits of implementing the idea.
We do our best to implement as many Ideas as we can. Our Product team will evaluate all submitted ideas in a timely manner and will disposition each into one of the following categories: will integrate into the product roadmap, further research is needed, unlikely to implement.
Thanks for collaborating with us!
Distributing a 1.4GB MSI to 15K servers presents some challenges and risks
The following two enhancements would be very beneficial to help better manage and throttle Chef system, network and target endpoint resources:
Reduce MSI package size. Currently 1.4GB
Separate Habitat client and Ruby integration within the same package, or remove dependency on Ruby.
We have encountered issues in the past where this can impact the network due to the package size. This would be a welcome change.
Attachments Open full size
I completely agree with this. Windows Chef-client has always been massive. Delivering it at scale, even in a non effortless manner, has always been a bit of an issue even more so in this method. We need to either find a way to remove the dependency if possible or, shrink it to bare minimum for client functionality
Attachments Open full size
We have 15,000 servers on a prod channel, if we push a new chef infra to prod, 15,000 servers are all gong to try to download a 1.4 GB package. If anything, separating the dependency of the ruby devkit from the install of Chef Infra would help so we only have to install the 1.4 GB Ruby devkit once, and not every time we do an upgrade to Chef Infra on Windows Servers.
Attachments Open full size