Cannot update fresh Fedora install (missing package from cisco)

A question: will the non-netinstaller ISO work? Does it include this file?

I understand your frustration but you have to understand this was purely a coincidence that you ran into this issue within the past few hours.

This is extremely fresh

Your point stands though, This shouldn’t stall the entire installation.

1 Like

I used full ISO here and it installed fine… just the first manual update afterwards fails.

You can ignore the failing rpm package, for the time being, with the --exclude command I shared above.

1 Like

There is a workaround for this issue, you can just disable that particular repo and the update/install will work fine. I have a feeling this will be rectified in due course at which time we can all re-enable the repo or add another one that will act as it’s mirror. And there are mirrors, but they look like they are all pointing to the same cluster… lol. Definitely speaks volumes to having a fallback server for cases like this, I guess

You can run this in your terminal to bypass the issue:

sudo dnf-3 config-manager --set-disabled fedora-cisco-openh264
1 Like

Thanks, I have worked around it already by installing Fedora without any Desktop environment and installing the package manually. It’s just funny because I was going to show people Fedora this week at work. We’re using Debian since forever, which I always found extremely stable, with official network mirrors even in my own city. But I’m challenging my own views and also wanted to challenge other people at work on why we still use Debian (the only drawback is that sometimes Debian uses old versions which requires more administrative work). If the full installer ISO doesn’t hold this package, I guess the problem will still show up…

Windows 11 is a nightmare, this is the time linux distros should focus on getting stability and usability for the desktop market… :slight_smile:

1 Like

to be fair, I have installed Fedora many times in the past and this is the first time it has ever caused any issues. I’ve even had more issues with Ubuntu than Fedora historically, which people don’t believe when I tell them usually, for it’s reputation for stability and supports anything out of the box.

Just goes to show that everyone has their own story to tell about this stuff I guess

3 Likes

I have installed fedora about half a year ago and had an issue where flatpaks would just crash for no reason(didn’t look much into it just jumped back to opensuse, I know not very wise of me), I have installed it again today and this happened hahah.

1 Like

somewhat semi-related… I was about to install proprietary/non-fee ‘RPM Fusion’ drivers and the NVidia driver… Is there a common/proper way to install all commonly used codecs/drivers/software that e.g. typically also come along by default with Ubuntu and the likes? and… will these h264 codecs/plugins and browser then also get (or need to be) updated to work with these changes?

It starts to feel a bit cumbersome to do this all manually … and I do get the whole concept of licenses and the desire to start with open-source as a base, but it’s kind of hard nowadays to get a modern PC working properly out of the box with mere open-source drivers…

Is there some tool/project/script that adds all the non-free/proprietary stuff to a fresh Fedora install?

(gotta hand it to Windows… it’s runs really well on a lottttt of systems… and to be fair… ever since 10, I have been using it (together with WSL or linux VMs) pretty reliably… )

I had this problem while installing chromium so unfortunately couldn’t disable the repo.

As @doineann said the trailing slash seems to be the issue so my workaround was manually downloading the package and then installing via the rpm:

wget -P /tmp  http://ciscobinary.openh264.org/openh264-2.4.1-2.fc41.x86_64.rpm
sudo dnf install /tmp/openh264-2.4.1-2.fc41.x86_64.rpm

Note the lack of a trailing slash in the url.

After that the install worked fine, this was for 41 (Container Image).

3 Likes

Just wondering… will this work well for later updates with the dnf package manager, when it gets fixed and the urls work again? or will a manual install mean that the package becomes ‘detached’ from dnf updates?

(I am kind new to Fedora and dnf… used it way back in the days of rpm and redhats in 90’s/00’s… so it’s all a bit new and different than the ‘usual’ Ubuntus for me)

Interesting that this worked for you. When I tried to this, it threw me the same error in the original post. I may have to check my syntax and the exact file that I grabbed with wget. Thanks for this suggestion!

That’s actually a good question. It could work either way. Does it search the repos for all existing packages and checks for updates, or does dnf record a package’s origin at the time of install and check against it’s source? In the case of some random rpm, it wouldn’t support updates, would it?

1 Like

I am no dnf/Fedora expert, but I can imagine that a manually installed rpm package (even when it is exactly the correct one, might still be ‘marked’ as a manual install and requires some special dnf command to be ‘marked for updates’ again.

Had problems like this with package managers in the past after manual updates…

Maybe there is someone here with more more knowledge who can elaborate on this?

I see that your original post is after ‘mozilla-openh264’ which wasn’t one of my dependancies but also seems to be failing. Try this:

wget -P /tmp http://ciscobinary.openh264.org/openh264-2.4.1-2.fc41.x86_64.rpm http://ciscobinary.openh264.org/mozilla-openh264-2.4.1-2.fc41.x86_64.rpm 
dnf install /tmp/openh264-2.4.1-2.fc41.x86_64.rpm /tmp/mozilla-openh264-2.4.1-2.fc41.x86_64.rpm

and then try running your update.

@doineann I was thinking just as a temporary fix, however I’m installing this stuff in containers I’m probably going to remove quite soon so it’s a bit different.

dnf info openh264 lists the repository as @commandline so I imagine dnf would have some trouble finding updates but I could be wrong.

Looks like it is currently being fixed :slight_smile: link

2 Likes

Thanks! exactly what I wondered/thought already… New things learned today!

So do I guess the correct and ‘proper’ way to do a manual install like @matt888 mentions, is to also set these ‘marked’ states correctly before and afterwards, and in the same way as originally done by the maintainers? I would already be confused as to what commands to use in this case.

as my intention is not to work temporarily with Fedora in a container, but have it as a main daily driver on my PC, I would prefer to do ‘fixes’ in a more ‘future proof’ way :wink:

Hi, I am from Brazil and I faced the same problem, but this suggestion worked very well for me, thanks.

1 Like

The issue was on the server side and it’s fixed now.

No need for workarounds anymore…

6 Likes

The repo, @commandline in this case, is info only and never used to find a newer rpm.

Ok interesting, so dnf will search for a package against any loaded repo and will store where the package was installed from for info only?

In terms of the above discussion this implies that as long as dnf has a correct package name then it will be able to search against loaded repos even if it was installed from a local rpm file?

I just did a little test by installing an old package and this looks to be the case:

wget -P /tmp https://fedora.mirror.digitalpacific.com.au/linux/releases/41/Everything/x86_64/os/Packages/o/openjpeg-2.5.2-4.fc41.x86_64.rpm
dnf install /tmp/openjpeg-2.5.2-4.fc41.x86_64.rpm
dnf info openjpeg # lists the newer package as available
dnf update # updates to the newer package
1 Like

Yes

Yes

You see this with the fedora-updates repos vs. the fedora repos for example.

1 Like