Red Hat Bugzilla – Bug 1005249
netinst with text mode doesn't allow to install GNOME (broken mirror?)
Last modified: 2014-04-07 11:01:14 EDT
Created attachment 794756 [details]
Description of problem:
If I boot netinst and don't touch anything at all, just refresh the screen, the metadata are downloaded from the closest mirror and than I see "[!] Software selection (Source changed - please verify)".
This is a recent change, it didn't use to force user to verify selection, if installation source wasn't changed (i.e. was on default).
Version-Release number of selected component (if applicable):
F20 Alpha TC4
Steps to Reproduce:
1. boot netinst
2. refresh screen, don't touch installation source
3. see "please verify" in software selection
user has to verify the default
user doesn't have to verify the default
Created attachment 794757 [details]
Created attachment 794758 [details]
Created attachment 794759 [details]
Created attachment 794760 [details]
Created attachment 794761 [details]
Created attachment 794762 [details]
Created attachment 794763 [details]
Actually, this is even funnier. I can't confirm that I want to install GNOME. If I enter the dialog and hit "c", I still see "please verify" error. If I pick KDE, everything seems OK. If I enter it again and change to GNOME, I again see "please verify" error. During screen refreshes I see a lot of yum errors.
This seems to be violating:
"When using the dedicated installer images, the installer must be able to use either HTTP or FTP repositories (or both) as package sources. The network install image must default to a valid publicly-accessible package source. "
Proposing as a blocker.
The same can be reproduced with DVD text mode. Just try to switch to Closest Mirror and you'll see the same behavior.
Also, the indicator next to Installation source if _very_ weird. Sometimes it displays [x] Closest Mirror. You refresh and you get [!] Closest Mirror. You refresh again and you get [x] Closest Mirror. Very confusing.
Samantha, are you aware of any recent changes that might have caused those issues?
(In reply to Vratislav Podzimek from comment #10)
> Samantha, are you aware of any recent changes that might have caused those
Maybe bug 1001081 is related? There were changes in installation source spoke.
(In reply to Kamil Páral from comment #11)
> (In reply to Vratislav Podzimek from comment #10)
> > Samantha, are you aware of any recent changes that might have caused those
> > issues?
> Maybe bug 1001081 is related? There were changes in installation source
No, there were no changes to the source spoke resulting from that bug--that was a fix for pykickstart. There have not been any recent changes which should be causing this at all, actually, but I'll look into it.
One recent change which does affect the '!' and 'x' (not complete/complete) markers beside each spoke went in on 05 Sept though, 77cdb0f07d179ca89db4e0ee339af761038bde44. The spoke was incorrectly being detected as completed, which could affect whether you are able to re-visit it.
Discussed at 2013-09-09 blocker review meeting . This may end up being a package/mirror issue and needs more data. will revisit at the next review meeting.
I have done a few retries and it worked for me now, both with netinst and DVD. I'll test more and comment here.
I have retried several times and I see no longer any problem (with TC4 or TC5). I'm withdrawing the blocker nomination. It was probably a broken mirror, but I'm not sure. Anaconda devs, can you find out the problem from the log files? Maybe anaconda could display a more helpful message in this case?
Closing this as NOTABUG since it was most likely a broken mirror.