[Spacewalk-list] Possible BUG: Requested packages have already been installed on CentOS 5 clients
Jan Hutař
jhutar at redhat.com
Tue Jan 26 21:09:31 UTC 2016
Thank you.
So far i have found something strange, but looks like only you
are hitting the issue. Could you please try this (on the client
which is failing to install the packages):
# python
Python 2.4.3 (#1, Oct 23 2012, 22:02:41)
[GCC 4.1.2 20080704 (Red Hat 4.1.2-54)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import yum.packages
>>> current = yum.packages.PackageEVR(None, '0.9.8e', '36.0.1.el5_11')
>>> new = yum.packages.PackageEVR(None, '0.9.8e', '37.el5_11')
>>> current.compare(new)
-1
>>> new = yum.packages.PackageEVR('', '0.9.8e', '37.el5_11')
>>> current.compare(new)
1
This is output I got on system with:
# rpm -q yum-rhn-plugin rpm rpm-python yum
yum-rhn-plugin-0.5.4.1-7.el5
rpm-4.4.2.3-36.el5_11
rpm-python-4.4.2.3-36.el5_11
yum-3.2.22-40.el5
(I know you have different yum-rhn-plugin version, but maybe
that is not important in this case)
After that, please add these 4 lines which will provide more
output to the /usr/share/rhn/actions/packages.py file after row
346:
--- usr/share/rhn/actions/packages.py.ORIG 2016-01-26 22:01:10.002192351 +0100
+++ usr/share/rhn/actions/packages.py 2016-01-26 22:04:56.623616475 +0100
@@ -344,6 +344,10 @@
found = False
for pkg in pkgs:
+ current = pkg.returnEVR()
+ print "Current:", current.epoch, current.version, current.release
+ print "Candidate:", evr.epoch, evr.version, evr.release
+ print "Comparasion:", current.compare(evr)
if pkg.returnEVR().compare(evr) == 0:
log.log_debug('Package %s already installed' \
% _yum_package_tup(package))
and regenerate `rhn_check -vv` output. Note I have not tried
this, so I'm not 100% sure it will work, so backup the file
first please.
And one more thing to be sure packages are OK:
rpm -qp --qf="%{EPOCH}:%{VERSION}-%{RELEASE}.%{ARCH}" openssl-0.9.8e-37.el5_11.i686.rpm
rpm -qp --qf="%{EPOCH}:%{VERSION}-%{RELEASE}.%{ARCH}" openssl-0.9.8e-36.0.1.el5_11.i686.rpm
or if you have them installed:
rpm -q --qf="%{EPOCH}:%{VERSION}-%{RELEASE}.%{ARCH}" openssl-0.9.8e-37.el5_11.i686
rpm -q --qf="%{EPOCH}:%{VERSION}-%{RELEASE}.%{ARCH}" openssl-0.9.8e-36.0.1.el5_11.i686
Regards,
Jan
On Tue, 26 Jan 2016 17:18:44 +0000 "Coffman, Anthony J"
<Tony.Coffman at snapon.com> wrote:
> Thanks for having a look.
>
> I should have also mentioned that a yum install or yum update
> does successfully install the same package(s) (from the same
> Spacewalk repos)
>
> Here's the rhn -vv output during an attempted update of
> openssl.
>
> https://paste.fedoraproject.org/314962/
>
> Here't the package list (rpm -qa | sort)
>
> https://paste.fedoraproject.org/314963/
>
> It does this if I try to update any package or groups of
> packages via scheduled action.
>
> Regards,
> --Tony
>
>
>
> -----Original Message-----
> From: Jan Hutař [mailto:jhutar at redhat.com]
> Sent: Monday, January 25, 2016 2:47 PM
> To: spacewalk-list at redhat.com
> Cc: Coffman, Anthony J <Tony.Coffman at snapon.com>
> Subject: Re: [Spacewalk-list] Possible BUG: Requested packages
> have already been installed on CentOS 5 clients
>
> On Mon, 18 Jan 2016 15:14:54 +0000 "Coffman, Anthony J"
> <Tony.Coffman at snapon.com> wrote:
>
> > I ran into this one yesterday unexpectedly.
> >
> > It seems there may be a bug in the SW 2.4 client that seems
> > to only affect EL5 clients. I tested this on CentOS 5 i386
> > but not on x86_64 yet. I haven't seen it on EL6, EL7, or
> > SuSE.
> >
> > When you scheduled a package install/update action, you get
> > this back
> >
> > Client execution returned "Requested packages already
> > installed" (code
> > 0)
> >
> > The 0 return code makes it look successful but no package
> > changes are made. Because of the change to spread out
> > errata cache processing over time you can easily be fooled
> > into thinking that you have successfully applied updates
> > when in reality nothing was actually applied.
> >
> > The requested packages are not actually already installed.
> >
> > I rolled one test system back to the 2.3 client manually and
> > it does fix the issue.
> >
> > Has anybody else run into this problem?
> >
> > --Tony
>
> As a first step, could you please run rhn_check on the client
> with "-vv" option and sent the full output? Also `rpm -qa |
> sort` might be handy.
>
> Regards,
> Jan
>
>
>
> --
> Jan Hutar Systems Management QA
> jhutar at redhat.com Red Hat, Inc.
--
Jan Hutar Systems Management QA
jhutar at redhat.com Red Hat, Inc.
More information about the Spacewalk-list
mailing list