Bug 486685

Can you try to update with selinux disabled? I think it has something with selinux, which can't be fixed in this release.

Try this:

state=`getenforce`
setenforce 0
yum update --enablerepo=updates-testing sagator-core
setenforce $state

Which version of sagator you have before update? (rpm -qa sagator\*)

There is no sagator-core in current fedora-10-stable. When trying to reproduce this on my machine (it's only a i386, but this should be similar for x86_64, because sagator is arch independent):

[root@localhost ~]# yum upgrade --enablerepo=updates-testing sagator-core
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * fedora: ftp.upjs.sk
 * updates-testing: ftp.upjs.sk
 * updates: ftp.upjs.sk
Setting up Upgrade Process
No Match for argument: sagator-core
Package(s) sagator-core available, but not installed.
No Packages marked for Update
[root@localhost ~]# 

Curious, that your errors are only on directories. May be it has something with fact, that these directories was not owned by sagator-1.1.0 and this bug has been fixed by sagator-1.1.1. I think I can do nothing with this error, because it's an error of sagator-1.1.0 and not this new version.

If you can help me to reproduce this bug on my machine, I can do more tests.

You are right about the command.  The two commands that actually ran:
yum -y update --enablerepo=updates-testing sagator-selinux.noarch 
yum -y update --enablerepo=updates-testing sagator-webq.noarch 

The sagator-core was pulled in as a dependency by yum.

Here is what is currently installed:

$ rpm -qa | grep -i sagator
sagator-selinux-1.1.1-0.beta1.fc10.noarch
sagator-1.1.1-1.fc10.noarch
sagator-webq-1.1.1-0.beta1.fc10.noarch

I haven't yet tried your workaround.  I'll update the ticket with those results later.

Thanks for your response.

Putting selinux in permissive mode allowed the transaction to succeed.

Dependency Installed:
  sagator-core.noarch 0:1.1.1-1.fc10

Updated:
  sagator-selinux.noarch 0:1.1.1-1.fc10                        sagator-webq.noarch 0:1.1.1-1.fc10

After the update there are the following packages installed:

$ rpm -qa | grep -i sagator
sagator-core-1.1.1-1.fc10.noarch
sagator-selinux-1.1.1-1.fc10.noarch
sagator-1.1.1-1.fc10.noarch
sagator-webq-1.1.1-1.fc10.noarch

There were *many* selinux errors indicated by SETroubleshooter during this update.

Well. I think this problems is caused by sagator-1.1.0 and already fixed in sagator-1.1.1, but I can't fix this before update, so this will fail if somebody has selinux enabled during update.

It looks, that selinux checks ownership of directories and directories, which are not owned by a package (they was missed in sagator<=1.1.0) are not permitted.
I think it's an feature of selinux policy and not a bug.

Can I close this bug as "CANTFIX"?

Fine by me.  What will happen when the updates get pushed to updates from updates-testing?  Will everyone who has selinux in enforcing mode then fail to update unless they turn enforcing off?  Seems you should ask someone for a workaround (Dan Walsh?) and rerelease with the fix.

But it is up to you.

I think he also can't do anything with this.

What happened after your fist update? sagator-1.1.1 was installed partially or not installed?

If installed partially, then may be only directory selinux contexts was not set properly, which is not critical.

I think it partially updated sagator-1.1.1.  

So that being the case, let's let it go. :-)