Bug 1001792

Summary: anaconda ppc64 "Please complete all spokes before continuing" with no way to complete Installation source
Product: [Fedora] Fedora Reporter: Richard W.M. Jones <rjones>
Component: anacondaAssignee: Anaconda Maintenance Team <anaconda-maint-list>
Status: CLOSED WONTFIX QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 19CC: anaconda-maint-list, dshea, g.kaviyarasu, jonathan, mkolman, nkadel, rjones, sbueno, vanmeeuwen+fedora
Target Milestone: ---   
Target Release: ---   
Hardware: ppc64   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2013-11-20 23:48:09 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Richard W.M. Jones 2013-08-27 18:53:28 UTC
Description of problem:

I'm trying to install Fedora 19 ppc64 on qemu.  See:
http://rwmj.wordpress.com/2013/08/27/booting-fedora-19-ppc64-netinst-under-qemu-on-x86-64/#content

This worked once, but now I cannot get anaconda to complete
the "spokes".  Particularly it's impossible to complete the
Installation Source (#1).  It just says either:

1) [!] Installation source
   (Processing ...)

which is understandable, or:

1) [!] Installation source
   (Closest mirror)

which is not.  If you try to go into 1, it just comes back to
the menu, sometimes after pausing for a few minutes (but with
no keyboard interaction).

Very very occasionally, you'll go into 1 and it will actually
offer the menu for installation source, let you select that
and so on.  However even on these rare occasions, it's still
showing [!] afterwards.

You cannot continue because it says:

Please complete all spokes before continuing

Version-Release number of selected component (if applicable):

Fedora 19 netinst ppc64 from secondary mirror.

How reproducible:

Uncertain, but apparently very often.

Steps to Reproduce:
1. Try to install.  See instructions in blog posting above.

Comment 1 David Shea 2013-08-27 18:57:42 UTC
Can you attach the log files from after a failure like this? What is the installation source status message after the failure to set it up?

Comment 2 Richard W.M. Jones 2013-08-27 21:35:12 UTC
I've deleted the VM now.  If you tell me precisely what
file to capture, I will try to run the test again.

The Installation Source status message was as described in comment 0.

Comment 3 Richard W.M. Jones 2013-08-28 11:32:50 UTC
Is it possible that you have to select the options
in numeric order?  I just did the install again and
selected option 1 first.  (Previously, since option 1
said "Processing ...", I was filling in the other options
before going back to option 1).

Comment 4 Richard W.M. Jones 2013-08-28 12:45:20 UTC
OK I believe I understand now what's going on here.

If I tail 'packaging.log' in another window, I can see that yum
is doing some sort of depsolving ("TSINFO: Marking <some package>
as install for <some other package>").  When that activity finishes
(which seems to take many minutes), spokes 1 and 6 are enabled.

Also every time you change the settings in spokes 1 or 6, it has
to go through this hidden depsolving process again.

This is all rather hidden and unexpected.  Couldn't it give
a bit more information about what's going on?  Even saying
"go to window 2 and tail -f /tmp/packaging.log" would be a
good start.

Comment 5 Samantha N. Bueno 2013-08-30 13:11:06 UTC
yum does the same depsolving in graphical installation--it's a bit more apparent there since the spoke status changes dynamically. Unfortunately that kind of dynamic status update is not possible with text mode.

We don't allow users to enter the Software Selection or Source Repo spoke in graphical installation if changes are made to either since the actions resultant from both are intertwined, and so downloading the yum metadata + depsolving must be complete if you wish to re-visit either spoke. The same constraints exist in graphical installation. Again, it is just more visible due to the dynamic status updates + being able to shade the icons when they are 'unavailable'.
The 'Processing... ' status under spokes in text UI indicates the spoke is unavailable visibly. It probably needs to be made more obvious or documented somewhere--but the 'r' key will redraw the screen, and any changes to the spoke status will be reflected then.

The log files can be found living in /tmp. tty2 should give you a shell where you can grab them.

Comment 6 Richard W.M. Jones 2013-11-21 09:55:59 UTC
Changing the resolution to indicate that you won't fix this,
but it is still a bug.

Comment 7 Nico Kadel-Garcia 2013-12-21 04:57:26 UTC
There is nothing "intertwining" about it. There is absolutely no point to doing component selections until after relevant upstream repositories are selected and enabled.. The older, non-spoke-and-wheel did this in order, and corectly.

If there are later steps to add additional repositories, then an instlalation user may have to back *up the ladder*, and repeat steps. But it's another loop, not an interweaving.

Please stop with this confusing model. It's a poor way to describe a limited state machine, one described by basic flow charts for the last 50 years of computing, and one that is actively being confused and harmed by the "spoke and wheel" paradigm.