Bug 602705

Additional info: 
The resulting anaconda-ks.cfg doesn't contain the --includepkgs option.

Does updates=http://clumens.fedorapeople.org/602705.img fix this?

The updates.img resulted in bug #607664

I've tested with snap #7 (0701.3 tree) and the ks.cfg line from comment #0 and I still don't see dracut package downloaded from the myrepo/ repository. access_log on the server shows that last files which were downloaded were:
/ks.cfg
/myrepo/.treeinfo
/myrepo/repodata/repomd.xml
/myrepo/repodata/primary.xml.gz

Anaconda log files don't make much sense for me. Let me know which file do you need. 

Moving back to ASSIGNED.

Hm, does it help if you also --excludepkgs= on the repo you don't want the package to come from?

I don't define this explicitly in the ks.cfg. This is the base repository which anaconda picks up when method=http://.

If I boot from the boot.iso with fully automated ks.cfg that has: 

repo --name="myrepo"  --baseurl=http://tyan-gt24-09.rhts.eng.bos.redhat.com/myrepo --includepkgs=dracut
repo --name=basic --baseurl=http://qafiler.bos.redhat.com/redhat/rel-eng/RHEL6.0-20100701.3/6/Server/x86_64/os/ --excludepkgs=dracut

i.e. define the base repostory then dracut is pulled from myrepo. 

A side question: How are --excludepkgs/--includepkgs supposed to work exactly?

This is starting to sound like something funny going on with yum, or at least something we need to be doing but aren't aware of.

                if ksrepo.excludepkgs:
                    repo.exclude = ksrepo.excludepkgs

                if ksrepo.includepkgs:
                    repo.include = ksrepo.includepkgs

...yes! We win the API confusion war!;).
 The internal API follows the yum config. names, so you have:

 [config]exclude     | --exclude | repo.exclode
 [config]includepkgs | N/A       | repo.includepkgs

James - but that should be fixed up in anaconda-13.21.55-1.

Alexander - can you please verify what version of anaconda you are testing?

Comment #5 is with anaconda-13.21.56-1 which is in the 0701.3 tree.

Looks like this is working in RHEL6.0-20100712.  At least, I'm unable to reproduce it.

Created attachment 433746 [details]
602705.cfg

so there are 2 repos with the exact same pkg in the repodata
yum won't treat them like mirrors - it will use ONE of the pkgs or the other.

so in the above case the failure to download it seems exactly correct, if it had chosen the pkg from the repodata that did not actually have the pkg FILE.

What's the status on this? Looks like anaconda and yum are doing the right thing but the feature still doesn't work in the way an end user will expect. 

How do we want this to work so QE can say this has been fixed or not?

The idea here is that you have multiple repos with some same package in them, and that you can use --includepkgs and --excludepkgs to tell kickstart/yum which repo should be used for those duplicate packages.  It's not intended to act as a mirror.

anaconda is just setting up some lists in repo objects, so I would assume that whatever's not happening here is in yum.  However if everything there looks fine and it's still not working, I am inclined to just close this one.  I expect that almost no one will care about these options in RHEL6.  This isn't worth the time being put into it.

Moving to VERIFIED per comment #21. We need to use --includepkgs and --excludepkgs on two separate repositories.

Red Hat Enterprise Linux 6.0 is now available and should resolve
the problem described in this bug report. This report is therefore being closed
with a resolution of CURRENTRELEASE. You may reopen this bug report if the
solution does not work for you.