In a bid to save download costs, we create a single Fedora repo.
In a bid to save (further) download costs establishing a Fedora 30 repo, is it possible to start from Fedora 29, rename the Fedora version component of the filenames (fc29 to fc30), and then use rsync update/delete to only download updated materials? (or is there some other aspect I’ve yet to realise)
Please advise =dn
I can’t help much, just one small thing to consider.
It won’t help to rename filenames from fc29 to fc30, as version (including .fc29 bit) is actually contained in the package metadata, so the same package files for F29 and F30 would still be different (whith bit-for-bit comparison).
On the other hand, F30 repo does actually contain some fc29 packages – quick and dirty look gives me number 1296 which may be wrong. Though it’s still a small percentage compared to the whole repo, and I’m not sure if the hassle of switching repo from fc29 to fc30 would justify the bandwidth saved.
I’ve never done local repo myself so can provide nothing that’s actually useful.
Thanks for the observation - at least I know the renaming effort to be a waste of time, if not bandwidth-saved.
In the same way that we are moving to system upgrades (cf rebuilding from scratch), maybe there should be something similar for repos?
(there are still many parts of the world (and outside of ‘well-connected’ country’s major cities) where bandwidth is expensive and transfer speeds slow!)
Yep, I’m totally with you on this one
Can you elaborate a bit?
Also do you already have a local repo for F29? Just curious.
Yes, we have a local repo for F29, just as we had for earlier releases. (am gradually building F30s, but it is taking time…)
Did you want me to elaborate on the repo-upgrade thought? It is possible to upgrade a single PC from F29 to F30 without stripping-back and installing the new OpSys from scratch/as a ‘green field’ operation. Accordingly, I had in mind a program which would go through the F29 repo ‘upgrading’ it to F30. This would only involve changing fileNM (my earlier thought?hope) and meta-data (your earlier contribution) in some cases, but would need new files/updated systems files to be downloaded/rsync-ed.
Valid or pipe-dream? =dn
As far as I understand, if you use a standard Fedora version upgrade process, you won’t be saving much bandwidth. Basically, for all the packages you’ve had installed in F29 it redownloads version for F30, then install F30 ones and remove F29 ones.
As far as I understand, it doesn’t save any bandwidth, just the time you spend configuring freshly installed system.
Or do you speak about some other Fedora version upgrade process, not
dnf system-upgrade one?
Yes, standard upgrade and update process.
Are you saying that the version upgrade process will not work from a local repo? Saving bandwidth is the primary reason for having local repos.
Am confused or have badly misunderstood… =dn
By that I meant standard process with remote repos, not local ones – and for one system, not many ones. Sorry to confuse you.
First of all, I really don’t know. I don’t understand the process well enough for definitive answer one way or another.
From what I do understand – some thoughts, but again, I’ve never done it and I don’t know for sure.
Technically it could work from local repo, unless fedora’s official repos are hardcoded in some way in
dnf system-upgrade. And I don’t think they should be hardcoded, it’s not logical… at least from the first glance ) Still it’s a thing to check.
Upgrade process may need both old repo and new repo to be available (to compare package lists, etc). Again, I’m not sure here, it can just take package names from installed system (not from the OLD repo), and search for them in the NEW repo. But if you already have the F29 repo, maybe you should keep a copy until you’re sure upgrade process won’t need it.