0

I have a server running CentOS (with Fasthosts UK).

Running a "yum upgrade" lists - and processes - a LOT of packages that it has upgrades for. Everything is downloaded and installed.

The problem is that if I do a subsequent "yum upgrade", it list all the same upgrades again! As if it was only running as a simulation and didn't really apply any upgrades.

This effectively means that I can't upgrade any of the software packages on my server ("yum upgrade specificpackage" also has the same problem) and so I wondered if anybody has any kind of experience or advice about this?

Is this a CentOS issue or a Fasthosts issue?

PS. FWIW there is also Plesk intalled on this server.

Improve this question
4
  • So you are running yum upgrade And then Typing 'y' for yes, and hitting enter?
    – Nathan Powell
    Jul 16, 2010 at 23:14
  • As I recall (have to admit I haven't done this v. recently) - yes
    – Dougal
    Jul 17, 2010 at 11:37
  • (348/348): java-1.6.0-openjdk-1.6.0.0-1.11.b16.el5.x86_64.rpm | 35 MB 00:04 ------------------------------------------------------------------------------------------------------- Total 2.0 MB/s | 360 MB 02:55 Running rpm_check_debug Running Transaction Test Finished Transaction Test Transaction Check Error: file /usr/bin/spfquery from install of perl-Mail-SPF-2.006-1.el5.rf.noarch conflicts with file from package spf2-1.2.5-0.274900.x86_64 Error Summary -------------
    – Dougal
    Jul 17, 2010 at 12:36
  • Above dump is poorly formatted, but a transaction test runs for a few mins (after all the downloads) then reports one transaction check error. no errors are listed in the error summary. But then doing a "yum upgrade" lists all the same package candidates again :-(
    – Dougal
    Jul 17, 2010 at 12:37

3 Answers 3

Reset to default
2

Based on that log snippet, it looks like you may have some repository priority conflicts. In my experience, the various repositories for CentOS are not well-coordinated. RPMForge doesn't always play nicely with EPEL, for example, and you get version number races.

To get more information, you could make yum run in verbose mode with the -d (debug) and -e (error level) switches. NUM should be between 1 and 10.

You might also try using --skip-broken, which should only hold back the package that's causing problems.

Improve this answer
4
  • Some v. useful info - thanks, I'll investigate. I have indeed added the RPMForge repo.
    – Dougal
    Jul 17, 2010 at 15:21
  • --skip-broken sounds great and I was actually imagining using something like that - will try - thanks
    – Dougal
    Jul 17, 2010 at 15:25
  • 1
    upboat - the problem is that yum is bailing out on you due to an error, so the right thing is actually happening. --skip-broken is your friend if you don't want to debug further, eventually it will get fixed upstream and work later (usually :) ).
    – user15590
    Jul 17, 2010 at 15:44
  • 1
    yum upgrade --disablerepo=rpmforge seems to be churning away now!
    – Dougal
    Jul 17, 2010 at 15:55
0

I haven't seen this specifically, but the first thing I would try is yum clean all to clean up package data, caches and the rpm database. Then see if you get the same results after another update.

Improve this answer
0

In your comments you say there is a "transaction check error". This is likely the problem. You should paste that here, although depending on what the problem was, it may have solved itself by now.

As other posters pointed out, some of the third party repos. are not very well managed. At most I would recommend: RHEL, EPEL, rpmfusion* and iuscommunity.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .