View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0000087||Rocky-Linux-8||anaconda||public||2021-10-14 18:35||2023-05-28 08:47|
|Reporter||Lukas Magauer||Assigned To||Infrastructure Team|
|Summary||0000087: Anaconda is unable to use mirrorlist or metalink as installation source on the first try|
|Description||While working on the test suites in OpenQA we discovered an issue with the Graphical Mirrorlist Installation.|
|Steps To Reproduce||- Start the graphical installer and select the language|
- Configure the network and clock
- Go in the "Installation Source" spoke and add either the [mirrorlist](mirrors.rockylinux.org/mirrorlist?repo=rocky-BaseOS-8.4&arch=x86_64) or the [metalink](mirrors.rockylinux.org/metalink?repo=rocky-BaseOS-8.4&arch=x86_64)
- Switch the URL type to the correct type
- Click done
After these steps the main screen will look like attachment_1
After another click on the "Install Source" spoke and going back with the done button, the mirrorlist/metalink gets loaded and accepted as source.
|Additional Information||## Expected behavior|
The mirrorlist/metalink gets directly recognized and used
## Further information found
- The problem looks to be that the BaseOS gets loaded by the mirrorlist/metalink but Anaconda is unable to find the linked AppStream repo (picture after this post)
- This problem does not occur if a repolist URL is used
I hope this is enough information, if something is missing or I can help with the troubleshooting, please just give me a note in MM.
|Tags||No tags attached.|
Install Source Error
Anaconda log from OpenQA with error
graphical-mirrorlist-installation-source-error_smaller.png (443,114 bytes)
I am unsure this is something that Release Engineering can fix. The .treeinfo is configured correctly as you can point to a direct mirror (such as dl.rl.o or anything that comes up in the list for your region) and it just works. The mirrorlist is behaving as it should as well, because dnf functions properly.
As this behavior *only* occurs while trying to use a mirror through anaconda, it will require Neil to assist looking into it, who is on vacation.
Okay I'm totally at you, this shouldn't be that high on the priority list, we thought maybe something for 8.5.
It just looked like a bug in Anaconda to us.
But maybe it is also just a configuration thing about the MirrorManager.
It could be either one. In 8.5 there is an anaconda version bump, so perhaps it has some fixes if it's not our mirror manager.
I had access to another 8.5-beta system with anaconda 188.8.131.52, and I know that there it still doesn't work. Same behavior as with our 8.4 image.
As this is kind of a minor problem, I set it to that now.
|Oh goody. Anaconda debugging!|
|This has been fixed a long time ago. (with the release of 8.6 I think)|