Bug 184426 - install error: unable to read package metadata
Summary: install error: unable to read package metadata
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: i386
OS: Linux
medium
high
Target Milestone: ---
Assignee: Anaconda Maintenance Team
QA Contact: Mike McLean
URL:
Whiteboard:
Keywords:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-03-08 17:44 UTC by Gianluca Cecchi
Modified: 2007-11-30 22:11 UTC (History)
1 user (show)

(edit)
Clone Of:
(edit)
Last Closed: 2006-03-08 19:04:25 UTC


Attachments (Terms of Use)

Description Gianluca Cecchi 2006-03-08 17:44:24 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:1.8.0.1) Gecko/20060111 Firefox/1.5.0.1

Description of problem:
I'm trying to install via pxe+nfs from rsync of today on x86 and I
get this error window during anaconda:
unable to read package metadata. this may be due
to a missing repodata. Please ensure that your
install tree has been correctly
generated. failure: repodata/comps.xml  from anaconda:
[errno 256] no more mirrors to try
switching to the shell with Ctrl+Alt+F2, I see inside anaconda.log in
/tmp I have
[snip]
10:42:24 INFO    : NFS install method detected, will use RHupdates/
10:42:24 INFO    : Running anaconda script /usr/bin/anaconda
10:42:27 INFO    : Display mode = g
10:42:27 INFO    : Method = nfs://mnt/source/.
10:42:44 INFO    : Started mini-wm
10:42:44 INFO    : Starting graphical installation...
10:42:45 INFO    : anaconda floppy device None
10:42:45 DEBUG   : self.driveList(): ['sda', 'sdb']
10:42:45 DEBUG   : DiskSet.skippedDisks: []
10:42:45 DEBUG   : DiskSet.skippedDisks: []
10:42:45 DEBUG   : starting all dmraids on drives ['sda', 'sdb']
10:42:45 DEBUG   : scanning for dmraid on drives ['sda', 'sdb']
10:42:45 WARNING : Couldnt lookup monitor type Smart Cable.
10:42:45 WARNING : step partitionmethod does not exist
10:42:45 WARNING : step partitionmethodsetup does not exist
10:42:45 WARNING : step autopartition does not exist
10:42:45 WARNING : step readcomps does not exist
10:42:45 WARNING : step selectlangpackages does not exist
10:42:45 WARNING : step handleX11pkgs does not exist
10:42:45 WARNING : step handlemiscpkgs does not exist
10:42:45 WARNING : step fixupconditionals does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:45 WARNING : step complete does not exist
10:42:46 INFO    : moving (1) to step partitionobjinit
10:42:46 DEBUG   : self.driveList(): ['sda', 'sdb']
10:42:46 DEBUG   : DiskSet.skippedDisks: []
10:42:46 DEBUG   : DiskSet.skippedDisks: []
10:42:46 INFO    : moving (1) to step autopartitionexecute
10:42:46 INFO    : moving (1) to step partitiondone
10:42:46 INFO    : moving (1) to step bootloadersetup
10:42:46 INFO    : moving (1) to step networkdevicecheck
10:42:46 INFO    : moving (1) to step reposetup
10:42:47 INFO    : anaconda
        [1/1]
10:42:47 ERROR   : reading package metadata: failure:
repodata/comps.xml from anaconda: [Errno 256] No more mirrors to try.

df -k gives:
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev                   1948452         0   1948452   0% /dev
192.168.0.110:/fc5nfs
                      14200000   9822368   3644672  73% /mnt/source
/tmp/loop0               70720     70720         0 100% /mnt/runtime
 and I can edit /mnt/source/repodata/comps.xml ....

Is this a problem on my part about rsync or what?
Same process is ok if using as nfs repository the dvd iso for test3, dumped to the directory.
I skip debug and SRPMS directory under tree for rsync but I think this doesn't matter...
How can I dig into this?

Version-Release number of selected component (if applicable):
anaconda-11.0.1-1

How reproducible:
Always

Steps to Reproduce:
1.setup pxe+dhcp+tftp server1
2.setup ks file on http server2
3. install server3 via pxe
  

Actual Results:  installation stops with the window message above and I can only exit the installation

Expected Results:  installation progress

Additional info:

Comment 1 Jeremy Katz 2006-03-08 18:11:54 UTC
Is your repodata properly synced?  Does the sha1sum of the comps file there
match that in repomd.xml

Comment 2 Gianluca Cecchi 2006-03-08 18:47:29 UTC
no,
[gcecchi@pevit00 fc5nfs]$ sha1sum repodata/comps.xml
a2f21fa1128ee7381d838216c310f371ac90caf6  repodata/comps.xml

inside repomd.xml instead I have:
  <data type="group">
    <location href="repodata/comps.xml"/>
    <checksum type="sha">424026225e0cea73daefa7fb7fe5623093b82b10</checksum>
    <timestamp>1139337044</timestamp>

while on dvd for test3
[gcecchi@pevit00 repodata]$ sha1sum comps.xml
dce353c5157588b329aa4e6b8975b9ee9b54ed34  comps.xml

and in repomd.xml:
    <location xml:base="media://1140120001.114809#1" href="repodata/comps.xml"/>
   <checksum type="sha">dce353c5157588b329aa4e6b8975b9ee9b54ed34</checksum>
    <timestamp>1140050497</timestamp>

so it seems also time stamp take role....
how to sync time stamps..? I used --size-only option in rsync command:

rsync -av --delete --size-only --exclude "debug/*"
rsync://mirrors.kernel.org/fedora/core/development/i386/ .

contents in rsynced dir are:
[gcecchi@pevit00 fc5nfs]$ ll repodata/
total 14356
-rw-r--r-- 1 gcecchi gcecchi  700983 Mar  8 08:27 comps.xml
-rw-r--r-- 1 gcecchi gcecchi 3623816 Mar  8 08:27 filelists.xml.gz
-rw-r--r-- 1 gcecchi gcecchi   28302 Mar  8 08:28 index.html
-rw-r--r-- 1 gcecchi gcecchi 8830850 Mar  8 08:27 other.xml.gz
-rw-r--r-- 1 gcecchi gcecchi 1266094 Mar  8 08:27 primary.xml.gz
-rw-r--r-- 1 gcecchi root       1140 Mar  8 08:27 repomd.xml
drwxr-xr-x 3 gcecchi root     204800 Mar  8 08:28 repoview

how can i see what means 1140050497 timestamp?

Comment 3 Warren Togami 2006-03-08 19:04:25 UTC
Please don't take this too harshly, but you are using our limited time with a
simple rsync usage problem.  For this reason I am closing NOTABUG.

However in my past experience running a Linux mirror, I believe I might be able
to help you.  Sometimes rsync even with seemingly proper options fails to have
an exact copy on the local side.  In those cases (like a corrupted package) I
would have to erase the local copy and run rsync again.  So try deleting that
stuff locally and rsync again, is it better?

I have no real explanation for why rsync sometimes fails in this way.  Maybe one
spot on a disk was going bad, and deleting the file and syncing again put the
file on a different physical spot on the disk so it worked around that problem.
 I don't know...


Comment 4 Gianluca Cecchi 2006-03-09 07:51:41 UTC
No problem with closing the issue. Sorry for your time stolen in this.
Release team requested to try devel kernels and installation kits, so I presumed
that also original mirror repodata dir download was involved in testing.
Running "createrepo -g repodata/comps.xml ." from nfs tree root, I can recreate
the repodata directory and install successfully from pxe+nfs, with updates at
rawhide report: 20060308 changes.
Probably you had better define two kinds of rsync mirrors:
- one for continuous updates
- one refreshed once a day with the directory tree suited for daily installation
only and during its own updates (its rsync client activity) its rsync server
services are stopped, so that if I try to rsync from my PC with this repository,
the tree is always consistent or I get an error because service is not available.

today, if I delete and re-sync the repodata dir, it seems I get consistent
results, thanks for the hint. 


Note You need to log in before you can comment on or make changes to this bug.