You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One issue we have is to roll out update between integration, staging and production.
It's difficult to perform a dnf (or yum in the past) update on integration to test the new packages, then to do the same on staging and than on production since the more time pass, the more updates are released.
I don't know the repository structure but in my mind I would do something like:
dnf update --> OUTPUT: ok done at this timestamp on integration
dnf update --timestamp on staging and production
The text was updated successfully, but these errors were encountered:
If I understand you correctly, you want DNF to be able to update to a state of packages as they were at the given timestamp in the repositories.
That's unfortunately impossible because of how the repositories are structured: A repository usually has a time stamp when it was generated, but once the repository is updated, the old state is lost and there is only the new timestamp and the new set of packages (those may and may not retain old packages). That means DNF has no way of retrieving the previous content of the repository, especially at an arbitrary timestamp.
You archive the installed file system into a tar ball and distributes the tar balls instead of separate packages. This is what containers (Docker, Podman) do.
You list exact versions of installed RPM packages (rpm -qa) on one system, copy the list to a target system and there you command DNF to install them (dnf distro-sync $(<FILE_WITH_THE_LIST)). That assumes that you keep all historical packages in the repository.
One issue we have is to roll out update between integration, staging and production.
It's difficult to perform a dnf (or yum in the past) update on integration to test the new packages, then to do the same on staging and than on production since the more time pass, the more updates are released.
I don't know the repository structure but in my mind I would do something like:
dnf update --> OUTPUT: ok done at this timestamp on integration
dnf update --timestamp on staging and production
The text was updated successfully, but these errors were encountered: