FC5T2 and Development issues, observations, and questions

[Nathan Grennan]

Rahul Sundaram wrote:
> Kindly enable SELinux and make sure it works rather than disabling it 
> during the test releases thereby ensuring that problems dont get 
> fixed. Reports bugs and post to this list if required. Firstboot stuff 
> is optional. So the installer becomes more streamlined. If you want to 
> change the developer's mind you need to provide detailed reasons.
  I have run into too many selinux only bugs to leave it enabled, 
especially with httpd. By it's very nature selinux goes against the KISS 
principle. I understand that part of the point of releasing and using 
test releases is for such testing, but why try to fix what is broken by 
design?
>>
>> 2. What is up with x86_64 kernels always being SMP enabled? I noticed 
>> a comment in the release notes, but no reason given as to why. I 
>> would think this could be problematic for debugging unless there is 
>> some boot option that makes the kernel act 100% like it would with a 
>> UP kernel.
>
> That is the intention. It doesnt make much sense anymore to run x86-64 
> system with a UP kernel. I will try and add a more detailed 
> explanation to the release notes. Remind me again after before test3 
> if you see problems.
I have since thought that users could just revert to i386 UP kernels if 
they had to, but on the flip side I have thought of situations were SMP 
only could be a real problem. One example would be a driver a user needs 
that is known to not play well with SMP. Their only recourse is to 
either wait till the driver is fixed, or revert to a i386 install.
>> 3. Where are the x86_64 Xen 3.0 kernels? All I see are i686 in 
>> development. I am really looking forward to playing with Xen 3.0 on 
>> my dual core x86_64 desktop.
>
> Work in progress to my understanding. see fedora-xen list. It is 
> scheduled to be available for FC5 afaik.
>
That is great to hear.
>>
>> 4. What is up with the return of mdadm.conf? I started with a FC4 
>> system with a /, /boot, and /home partition. I told anaconda to 
>> format / and /boot, while ignoring /home. I told it to ignore home 
>> because that is where I backed up all the data. I then finished 
>> installing FC5T2. After booting for the first time I setup /home in 
>> fstab, it is /dev/md2. I then rebooted and got an error about 
>> /dev/md2. After digging into the problem I found udev was not 
>> creating the file /dev/md2. I thought of mdadm.conf from the past, 
>> and looked in it. Sure enough it existed and mentioned md0 and md1. I 
>> added md2, and then /home would mount properly on boot. I thought we 
>> had gotten rid of the need for mdadm.conf.
>
> File a bug report.
>
  Ok, I expected the a developer to come back with an excuse similar to 
the one I have gotten about using partition labels to mount.
>
>>
>> 10. Anaconda's dependency code may be a little screwy. I am pretty 
>> sure I had told it to install beagle during the package selection. I 
>> think I deselect one of it's dependencies. Instead of installing the 
>> dependency anyway so that it could install beagle, it seemed to not 
>> install either. Another possibility is that the package selection 
>> code is just buggy. I did notice that the selection display was new. 
>> I am happy to see it's return after the lack of a decent one in FC5T1.
>
> Can you file a bug report against anaconda with the installation logs 
> in /root?
>
Yes, I should still have those logs in /root.