On Tue, Jan 22, 2013 at 9:15 PM, Johnny Hughes johnny@centos.org wrote:

On 01/22/2013 03:01 PM, James Freer wrote:

...

I've just installed v6.3 as a desktop (from Centos-6.3-i386-LiveCD.iso) and to get the hang of the Centos approach and then hope to move on to a server. I've been using linux *buntu for 5 years.

Hope i don't sound like a nit but i've got a little confused with the repos. Hoping someone would be kind enough just to clarify. This installation is for stability whilst installing the latest versions available.

a] What happens if i select these? I assume that Centos6 is correct so why these?

In software sources i have checked; Centos6-base, Centos6-extras, Centos6-updates

unchecked; Centos6-contrib, Centos6-media, Centos6-plus i did check these three and media wouldn't, and plus wanted to suggest installation kernel 2.6 [gulp]. Each unchecked without installing.

Then we also have the same for 6.0, 6.1, 6.2?

b] So which repos to use as i'm unsure where apps are i want to use like efax (ok that's in), yumex, pyrenamer, llgal, abiword. at present i'm not sure which repos to use other than the one's i've mentioned.

I then started to look at the repos http://wiki.centos.org/AdditionalResources/Repositories

Centos extras and plus -- are these the same as above? CS/GFS, Centos testing, Fast track, debuginfo, contrib, CR - would seem not applicable to me.

3rd party repos rpmforge (DAG), and EPEL seem likely other repos seem likely to replace core packages which wouldn't be applicable. Google RPM gets a "use with care". I used the google browser on *buntu from their website with no problems so i was wondering what the problem would be and whether to use this repoo or the website. The fact that a maintainer has made this repo suggests a reason rather than just convenience.

c] Could one just use pkgs.org to install from?

My google "centos abiword" threw up yet another repo 'puias' and this has the latest Abiword version 2.8.6. http://pkgs.org/centos-6-rhel-6/puias-i386/abiword-2.8.6-3.puias6.i686.rpm.h... I could just install from pkgs.org/centos-6 it would seem? Point is the repos are maintained by an expert rather than folk just downloading.

I was then confused further by this http://miles.isb.bj.edu.cn/2012/03/09/how-to-install-abiword-on-centos-6-x64... One needs to add epal and rpmforge just for abiword? Surely the dependencies would be in the same repo

I'm sure all is quite simple really but i'd be grateful for the guidance before i mess up an installation.

puias is another rebuild of RHEL code (like CentOS .. but a different project) ... they also just changed their name to Springdale Linux:

http://springdale.math.ias.edu/

WRT recommended repos (that also explains what extras is, centosplus is, etc):

http://wiki.centos.org/AdditionalResources/Repositories

Basically, for almost everything I need, I can get it from CentOS, EPEL, and maybe RPMForge.

As far as the distro is concerned, CentOS-6 is the distro and the point releases are basically just point in time freezes to generate new install media. You will always get to the latest version of CentOS-6 (currently 6.3 and updates to that) by doing "yum update"

CR is a way to "point release" updates about a week faster, we release the RPMS when they are ready, into CR and then we make ISOs out of them and release the next version. It usually takes 5-10 days to make the ISOs, test them and seed them.

puias... sorry i overlooked that. Clear enough now. I'll use the epal and rpmforge repos and see how i get on.

thanks james

On Wed, Jan 23, 2013 at 10:15 AM, Leon Fauster leonfauster@googlemail.com wrote:

Am 22.01.2013 um 23:03 schrieb James Freer jessejazza3.uk@gmail.com:

...

On Tue, Jan 22, 2013 at 9:15 PM, Johnny Hughes johnny@centos.org wrote:

...

On 01/22/2013 03:01 PM, James Freer wrote:

...

I've just installed v6.3 as a desktop (from Centos-6.3-i386-LiveCD.iso) and to get the hang of the Centos approach and then hope to move on to a server. I've been using linux *buntu for 5 years.

Hope i don't sound like a nit but i've got a little confused with the repos. Hoping someone would be kind enough just to clarify. This installation is for stability whilst installing the latest versions available.

a] What happens if i select these? I assume that Centos6 is correct so why these?

In software sources i have checked; Centos6-base, Centos6-extras, Centos6-updates

unchecked; Centos6-contrib, Centos6-media, Centos6-plus i did check these three and media wouldn't, and plus wanted to suggest installation kernel 2.6 [gulp]. Each unchecked without installing.

Then we also have the same for 6.0, 6.1, 6.2?

b] So which repos to use as i'm unsure where apps are i want to use like efax (ok that's in), yumex, pyrenamer, llgal, abiword. at present i'm not sure which repos to use other than the one's i've mentioned.

I then started to look at the repos http://wiki.centos.org/AdditionalResources/Repositories

Centos extras and plus -- are these the same as above? CS/GFS, Centos testing, Fast track, debuginfo, contrib, CR - would seem not applicable to me.

3rd party repos rpmforge (DAG), and EPEL seem likely other repos seem likely to replace core packages which wouldn't be applicable. Google RPM gets a "use with care". I used the google browser on *buntu from their website with no problems so i was wondering what the problem would be and whether to use this repoo or the website. The fact that a maintainer has made this repo suggests a reason rather than just convenience.

c] Could one just use pkgs.org to install from?

My google "centos abiword" threw up yet another repo 'puias' and this has the latest Abiword version 2.8.6. http://pkgs.org/centos-6-rhel-6/puias-i386/abiword-2.8.6-3.puias6.i686.rpm.h... I could just install from pkgs.org/centos-6 it would seem? Point is the repos are maintained by an expert rather than folk just downloading.

I was then confused further by this http://miles.isb.bj.edu.cn/2012/03/09/how-to-install-abiword-on-centos-6-x64... One needs to add epal and rpmforge just for abiword? Surely the dependencies would be in the same repo

I'm sure all is quite simple really but i'd be grateful for the guidance before i mess up an installation.

puias is another rebuild of RHEL code (like CentOS .. but a different project) ... they also just changed their name to Springdale Linux:

http://springdale.math.ias.edu/

WRT recommended repos (that also explains what extras is, centosplus is, etc):

http://wiki.centos.org/AdditionalResources/Repositories

Basically, for almost everything I need, I can get it from CentOS, EPEL, and maybe RPMForge.

As far as the distro is concerned, CentOS-6 is the distro and the point releases are basically just point in time freezes to generate new install media. You will always get to the latest version of CentOS-6 (currently 6.3 and updates to that) by doing "yum update"

CR is a way to "point release" updates about a week faster, we release the RPMS when they are ready, into CR and then we make ISOs out of them and release the next version. It usually takes 5-10 days to make the ISOs, test them and seed them.

puias... sorry i overlooked that. Clear enough now. I'll use the epel and rpmforge repos and see how i get on.

Hi folks

i do not recommend to use this two repos simultaniuous (or use priorities!). LF

hmmm - well what is one supposed to do? I've got EPEL installed fine but that doesn't have abiword, pyrenamer and some of the the other... fairly standard apps. I turned to Centos after Fedora has proved to be a bit dodgy. Centos along with debian are supposed to be the TWO main community distros. If rpmforge shouldn't be installed (as you've advised) it seems i need to look at another distro. The advice is to stay with a distro's developer packages and only use other repo if one really has to. I can only assume that Centos really is for Server use rather than desktop... i was just hoping to use it as a desktop before moving onto the server route.

james

On 1/24/2013 3:27 PM, James Freer wrote:

On Wed, Jan 23, 2013 at 10:15 AM, Leon Fauster leonfauster@googlemail.com wrote:

...

Am 22.01.2013 um 23:03 schrieb James Freer jessejazza3.uk@gmail.com:

...

On Tue, Jan 22, 2013 at 9:15 PM, Johnny Hughes johnny@centos.org wrote:

...

On 01/22/2013 03:01 PM, James Freer wrote:

...

I've just installed v6.3 as a desktop (from Centos-6.3-i386-LiveCD.iso) and to get the hang of the Centos approach and then hope to move on to a server. I've been using linux *buntu for 5 years.

Hope i don't sound like a nit but i've got a little confused with the repos. Hoping someone would be kind enough just to clarify. This installation is for stability whilst installing the latest versions available.

a] What happens if i select these? I assume that Centos6 is correct so why these?

In software sources i have checked; Centos6-base, Centos6-extras, Centos6-updates

unchecked; Centos6-contrib, Centos6-media, Centos6-plus i did check these three and media wouldn't, and plus wanted to suggest installation kernel 2.6 [gulp]. Each unchecked without installing.

Then we also have the same for 6.0, 6.1, 6.2?

b] So which repos to use as i'm unsure where apps are i want to use like efax (ok that's in), yumex, pyrenamer, llgal, abiword. at present i'm not sure which repos to use other than the one's i've mentioned.

I then started to look at the repos http://wiki.centos.org/AdditionalResources/Repositories

Centos extras and plus -- are these the same as above? CS/GFS, Centos testing, Fast track, debuginfo, contrib, CR - would seem not applicable to me.

3rd party repos rpmforge (DAG), and EPEL seem likely other repos seem likely to replace core packages which wouldn't be applicable. Google RPM gets a "use with care". I used the google browser on *buntu from their website with no problems so i was wondering what the problem would be and whether to use this repoo or the website. The fact that a maintainer has made this repo suggests a reason rather than just convenience.

c] Could one just use pkgs.org to install from?

My google "centos abiword" threw up yet another repo 'puias' and this has the latest Abiword version 2.8.6. http://pkgs.org/centos-6-rhel-6/puias-i386/abiword-2.8.6-3.puias6.i686.rpm.h... I could just install from pkgs.org/centos-6 it would seem? Point is the repos are maintained by an expert rather than folk just downloading.

I was then confused further by this http://miles.isb.bj.edu.cn/2012/03/09/how-to-install-abiword-on-centos-6-x64... One needs to add epal and rpmforge just for abiword? Surely the dependencies would be in the same repo

I'm sure all is quite simple really but i'd be grateful for the guidance before i mess up an installation.

puias is another rebuild of RHEL code (like CentOS .. but a different project) ... they also just changed their name to Springdale Linux:

http://springdale.math.ias.edu/

WRT recommended repos (that also explains what extras is, centosplus is, etc):

http://wiki.centos.org/AdditionalResources/Repositories

Basically, for almost everything I need, I can get it from CentOS, EPEL, and maybe RPMForge.

As far as the distro is concerned, CentOS-6 is the distro and the point releases are basically just point in time freezes to generate new install media. You will always get to the latest version of CentOS-6 (currently 6.3 and updates to that) by doing "yum update"

CR is a way to "point release" updates about a week faster, we release the RPMS when they are ready, into CR and then we make ISOs out of them and release the next version. It usually takes 5-10 days to make the ISOs, test them and seed them.

puias... sorry i overlooked that. Clear enough now. I'll use the epel and rpmforge repos and see how i get on.

Hi folks

...

i do not recommend to use this two repos simultaniuous (or use priorities!). LF

hmmm - well what is one supposed to do? I've got EPEL installed fine but that doesn't have abiword, pyrenamer and some of the the other... fairly standard apps. I turned to Centos after Fedora has proved to be a bit dodgy. Centos along with debian are supposed to be the TWO main community distros. If rpmforge shouldn't be installed (as you've advised) it seems i need to look at another distro. The advice is to stay with a distro's developer packages and only use other repo if one really has to. I can only assume that Centos really is for Server use rather than desktop... i was just hoping to use it as a desktop before moving onto the server route.

james _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

I use some things from both epel and rpmforge on many systems. However, I have used the yum confs to enable what I need from one and exclude everything else. Then if I'm worried about the other having what I'm pulling in from the one, I disable that in its conf file. You should read up on the power of yum where you can set priorities, do exclusions and such.

Yes, you can get into trouble if you add 2 without any control. For instance, something like clamav. One repo might set it up with the username of clam while the other might use clamav. As the updates come down, suddenly it dies and you have to figure out that the logs are owned by the wrong user. This is just one example of many things that can go wrong with mixed repos.

On 24.01.2013 20:27, James Freer wrote:

On Wed, Jan 23, 2013 at 10:15 AM, Leon Fauster leonfauster@googlemail.com wrote:

...

Am 22.01.2013 um 23:03 schrieb James Freer jessejazza3.uk@gmail.com:

...

On Tue, Jan 22, 2013 at 9:15 PM, Johnny Hughes johnny@centos.org wrote:

...

On 01/22/2013 03:01 PM, James Freer wrote:

...

I've just installed v6.3 as a desktop (from Centos-6.3-i386-LiveCD.iso) and to get the hang of the Centos approach and then hope to move on to a server. I've been using linux *buntu for 5 years.

Hope i don't sound like a nit but i've got a little confused with the repos. Hoping someone would be kind enough just to clarify. This installation is for stability whilst installing the latest versions available.

a] What happens if i select these? I assume that Centos6 is correct so why these?

In software sources i have checked; Centos6-base, Centos6-extras, Centos6-updates

unchecked; Centos6-contrib, Centos6-media, Centos6-plus i did check these three and media wouldn't, and plus wanted to suggest installation kernel 2.6 [gulp]. Each unchecked without installing.

Then we also have the same for 6.0, 6.1, 6.2?

b] So which repos to use as i'm unsure where apps are i want to use like efax (ok that's in), yumex, pyrenamer, llgal, abiword. at present i'm not sure which repos to use other than the one's i've mentioned.

I then started to look at the repos http://wiki.centos.org/AdditionalResources/Repositories

Centos extras and plus -- are these the same as above? CS/GFS, Centos testing, Fast track, debuginfo, contrib, CR - would seem not applicable to me.

3rd party repos rpmforge (DAG), and EPEL seem likely other repos seem likely to replace core packages which wouldn't be applicable. Google RPM gets a "use with care". I used the google browser on *buntu from their website with no problems so i was wondering what the problem would be and whether to use this repoo or the website. The fact that a maintainer has made this repo suggests a reason rather than just convenience.

c] Could one just use pkgs.org to install from?

My google "centos abiword" threw up yet another repo 'puias' and this has the latest Abiword version 2.8.6. http://pkgs.org/centos-6-rhel-6/puias-i386/abiword-2.8.6-3.puias6.i686.rpm.h... I could just install from pkgs.org/centos-6 it would seem? Point is the repos are maintained by an expert rather than folk just downloading.

I was then confused further by this http://miles.isb.bj.edu.cn/2012/03/09/how-to-install-abiword-on-centos-6-x64... One needs to add epal and rpmforge just for abiword? Surely the dependencies would be in the same repo

I'm sure all is quite simple really but i'd be grateful for the guidance before i mess up an installation.

puias is another rebuild of RHEL code (like CentOS .. but a different project) ... they also just changed their name to Springdale Linux:

http://springdale.math.ias.edu/

WRT recommended repos (that also explains what extras is, centosplus is, etc):

http://wiki.centos.org/AdditionalResources/Repositories

Basically, for almost everything I need, I can get it from CentOS, EPEL, and maybe RPMForge.

As far as the distro is concerned, CentOS-6 is the distro and the point releases are basically just point in time freezes to generate new install media. You will always get to the latest version of CentOS-6 (currently 6.3 and updates to that) by doing "yum update"

CR is a way to "point release" updates about a week faster, we release the RPMS when they are ready, into CR and then we make ISOs out of them and release the next version. It usually takes 5-10 days to make the ISOs, test them and seed them.

puias... sorry i overlooked that. Clear enough now. I'll use the epel and rpmforge repos and see how i get on.

Hi folks

...

i do not recommend to use this two repos simultaniuous (or use priorities!). LF

hmmm - well what is one supposed to do? I've got EPEL installed fine but that doesn't have abiword, pyrenamer and some of the the other... fairly standard apps. I turned to Centos after Fedora has proved to be a bit dodgy. Centos along with debian are supposed to be the TWO main community distros. If rpmforge shouldn't be installed (as you've advised) it seems i need to look at another distro. The advice is to stay with a distro's developer packages and only use other repo if one really has to. I can only assume that Centos really is for Server use rather than desktop... i was just hoping to use it as a desktop before moving onto the server route.

If you'd like to try, I do maintain a desktop related repository meant to not conflict with @Base or EPEL (or Elrepo): rpm -ivh http://li.nux.ro/download/nux/dextop/el6/x86_64/nux-dextop-release-0-2.el6.n...

There are a lot of packages there, for those that are not you can drop me an email or ask on https://forums.nux.ro/index.php?t=thread&frm_id=10&

If you'd like to get even more adventurous I even have a desktop Centos remix built around EPEL, ElRepo and my own repos: http://li.nux.ro/stella/

</shameless plug>

From: James Freer jessejazza3.uk@gmail.com

...

i do not recommend to use this two repos simultaniuous (or use priorities!).

hmmm - well what is one supposed to do?

I use priorities and did not run into much problems so far... base = 1 rpmforge = 2 (just watch out for rfx packages) elrepo = 3 epel = 4

JD

On 25.01.2013 20:15, James Freer wrote:

Thanks for that suggestion. I'll look into it all again in a couple of days. I'm just surprised that Centos risk the stability issue using other repos.

This goes for any distro; once you start adding 3rd party stuff you get further and further from the promise of stability.

On Fri, Jan 25, 2013 at 2:32 PM, m.roth@5-cent.us wrote:

...

I wouldn't agree with the priorities above. As I mentioned earlier in this

thread, our standard repos, in addition to the base, are rpmfusion and epel. Other than *sigh* 386 vs x86_64 issues, we've never had dependency conflicts. It all just works. elrepo I use *only* for kmod-nvidia and its dependencies; I don't use rpmforge - I've tried, a few times, and frequently run into dependency conflicts.

Have you used rpmforge since the split into conflicting and non-conflicting sections?

On Fri, Jan 25, 2013 at 3:13 PM, m.roth@5-cent.us wrote:

...

...

Have you used rpmforge since the split into conflicting and non-conflicting sections?

Didn't know they had...

I think it was a couple of years ago. See the note here about rpmforge-extras: http://wiki.centos.org/AdditionalResources/Repositories/RPMForge

You can still run into some quirks if the same package appears in epel or centos-extras since those are't considered 'base',

James Freer wrote:

On Fri, Jan 25, 2013 at 8:32 PM, m.roth@5-cent.us wrote:

...

James Freer wrote:

...

On Fri, Jan 25, 2013 at 1:34 PM, John Doe jdmls@yahoo.com wrote:

...

From: James Freer jessejazza3.uk@gmail.com

...

...

i do not recommend to use this two repos simultaniuous (or use priorities!).

hmmm - well what is one supposed to do?

I use priorities and did not run into much problems so far... base = 1 rpmforge = 2 (just watch out for rfx packages) elrepo = 3 epel = 4

JD

Thanks for that suggestion. I'll look into it all again in a couple of days. I'm just surprised that Centos risk the stability issue using other repos.

I wouldn't agree with the priorities above. As I mentioned earlier in this thread, our standard repos, in addition to the base, are rpmfusion

and

...

epel. Other than *sigh* 386 vs x86_64 issues, we've never had dependency conflicts. It all just works. elrepo I use *only* for kmod-nvidia and its dependencies; I don't use rpmforge - I've tried, a few times, and frequently run into dependency conflicts.

I was only thinking of rpmforge as it seems the only repo with Abiword in it. For WP one has only got LO-writer or Abiword... i prefer the lighter unless i need the features of LO-writer as i did when i was teaching and preparing handouts.

Watch out how you say that - my reaction on seeing you write WP is to wonder if they've *finally* put out a newer release of WordPerfect. (There *was* a linux version about 10-12 years ago, under Corel Linux).

mark

On 1/26/2013 2:06 PM, Mark LaPierre wrote:

There was a WordPerfect included in Red Hat Linux around about release 5 or so. To get it to work on later releases you had to install the required dependency packages from the install CD along with the WordPerfect package. I bet it would still run on CentOS if you installed it with it's dependency packages.

not if those dependencies, like an ancient version of libc, libX, etc, conflict with the much much newer versions in your RHEL/CentOS version.

On 26.01.2013 22:51, James Freer wrote:

Not sure how word perfect got into the thread. I used the letters WP for word processing and now Word Perfect is being discussed

LOL! Legendary ... I love this mailing list!

Am 25.01.2013 um 21:15 schrieb James Freer jessejazza3.uk@gmail.com:

On Fri, Jan 25, 2013 at 1:34 PM, John Doe jdmls@yahoo.com wrote:

...

From: James Freer jessejazza3.uk@gmail.com

...

hmmm - well what is one supposed to do?

I use priorities and did not run into much problems so far... base = 1 rpmforge = 2 (just watch out for rfx packages) elrepo = 3 epel = 4

JD

Thanks for that suggestion. I'll look into it all again in a couple of days. I'm just surprised that Centos risk the stability issue using other repos.

not the CentOS(-Team) but the user it self is risking this …

-- LF

Nikolaos Milas wrote:

On 25/1/2013 11:28 μμ, Leon Fauster wrote:

...

not the CentOS(-Team) but the user it self is risking this …

<snip>

You must balance availability of features, stability, manageability, security, package dependencies, application/service deployment and maintenance and more.

<snip>

As a simple example, if there is a requirement to run OpenLDAP *as a server* on a CentOS OS, the sysadmin MUST find *how* to run the latest version (which is the only "approved" one for OpenLDAP server deployments by the OpenLDAP project). Deploying OpenLDAP using the packages available by either CentOS 5 or 6 repos is unacceptable.

Ah, but "the latest" is a) a moving target, and b) not what I would ever use in a development or production environment, nor at home, unless I really did want to spend time debugging the system. In fact, I'd normally *NOT* ever install an x.0 release - I always wait for at least x.0.1 or x.1, so that all the bugs found by the early adopters have mostly been squashed.

mark

Received from Nikolaos Milas, on 2013-01-25 10:05 PM:

On 25/1/2013 11:28 μμ, Leon Fauster wrote:

...

not the CentOS(-Team) but the user it self is risking this …

True. CentOS/RHEL are using the least-risk policy by rarely updating packages, except for serious bug/security fixes and that helps provide peace of mind from the base OS.

Yet, I have come to believe that Systems Administration is not trivial in terms of decision-making; in fact, one could say that it may be a highly philosophical (!) job.

You must balance availability of features, stability, manageability, security, package dependencies, application/service deployment and maintenance and more.

Experience, knowledge and a thoughtful attitude will hopefully help find a "golden section" between all these through time on a per case-basis.

No systems are identical. The sysadmins have to *study* their environment and needs and then design the proper solution on each case.

As a simple example, if there is a requirement to run OpenLDAP *as a server* on a CentOS OS, the sysadmin MUST find *how* to run the latest version (which is the only "approved" one for OpenLDAP server deployments by the OpenLDAP project). Deploying OpenLDAP using the packages available by either CentOS 5 or 6 repos is unacceptable.

2c, Nick _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

Thanks Nikolaos/Nick. Very accurate, from my applied case/scenario as well.

Different demand and functionality requirements causes different types of settings for different repos.

And on some boxes where i want to deploy/apply/expect latest apps and services or functionality, (like one of them you've mentioned is OpenLDAP), CentOS requires me to do massive modifications.

I find it very very annoying that, CentOS lacks STABLE+last releases. It is not only CentOS, ther Linux as well. But this RHEL close/derivative, is very very behind.

I needed & still need to do vast amount of includepkgs=... exclude=... priority=... protect=... etc very careful implementation & management on all .repo files. And on top of that, YUM, understands only (the last) one line of "includepkgs=..", "protect=..". :-(

If it(YUM) were to understand multiple of those config options/lines, then, different type/category of apps/tools could have been added, copy/pasted or moved/placed easily on different channels of different related repo, based on their "priority" sequence.

And YUM need to have a feature to analyze a user specified/given app. IF, yum were to have a feature to analyze current priority, include, exclude settings, and then show/indicate what include, exclude need to be set for an user-specified or pre-specified last+stable app/tool, then such would have been very helpful. Yum need to analyze all deps/libs related to that pre-specified app.

And, may be even a better chroot type of app/system should be developed & exist in CentOS/RHEL, to easily try those STABLE+last releases, effectively, so that service based on those can be easily used, even on a 128 MB based box.

CentOS webpage/site should also show to all users, some example of using multiple repos and how to implement effective includepkgs, exclude, priority etc directives properly for some certain last & STABLE app(s) (which is by default not in CentOS), so that others can understand the pattern, or have a pointer for them.

Just mentioning about, that, there is such things called "includepkgs=...", "exclude=..." ad now go do it yourself (and sorry no example), obviously does not help that much to users, and its CentOS's loss as well, users go away to other distros, and ultimately many of them are lost in the jungle.

-- Bright Star (Bry8Star).

On 26/1/2013 7:42 πμ, Les Mikesell wrote:

If you find a 100% reliable solution, please post it.

I suggest very carefully selecting particular packages (and groups thereof) from whatever repos (or from individual experienced users' efforts), testing them thoroughly on test systems (usually a set of VMs) and/or building your own packages based on selected/modified SRPMs (according to your own requirements), then defining your own enterprise repo(s) for internal use.

Why not use other distro(s) with more recent packages? Because I have tried them and I only find RHEL / CentOS a truly Enterprise System. At least it suits me. :-)

There is no such thing as a 100% reliable system if you don't do things right. You can easily screw (in terms of security and reliability) your Vanilla RHEL/CentOS with one wrong setting, even using your safe repos.

Nick

On Sat, Jan 26, 2013 at 4:22 PM, Les Mikesell lesmikesell@gmail.com wrote:

On Sat, Jan 26, 2013 at 1:44 AM, Nikolaos Milas nmilas@noa.gr wrote:

...

...

There is no such thing as a 100% reliable system if you don't do things right. You can easily screw (in terms of security and reliability) your Vanilla RHEL/CentOS with one wrong setting, even using your safe repos.

Breaking it yourself is a different issue than installing conflicting things from uncoordinated repositories though. The underlying problem is that using newer packages often requires replacing core libraries. Any single 3rd party repo doing this will probably test things together, but if you combine 2 or more of these you are mixing untested code, likely to have had different compile options, etc. This seems especially bad in the RH/CentOS world because the core repos and epel have very restrictive policies on what can be there and the 3rd party repos are intentionally uncoordinated. I think the debian/Ubuntu world has many more packages in at least loosely coordinated repos. You can still run into problems but you are less likely to be the first person to every mix a particular set of packages in an install. Les Mikesell

From what i have seen of fedora and centos in the rpm world the repos

are very much better in the debian world. To me the stability comes from the distro and it's repos. Not being able to install Abiword or yumex, having to spend time selecting options for repos to me simply isn't worth it.

I've just installed a Slackware distro today and it's the best i've ever tried in 6 years of using linux. It's speed, ease of installation put's it in a league of its own. Or as their 'chilling warning goes' Once you go Slack... you never go back!

james

On 26/01/13 14:59, Bry8 Star wrote:

CentOS webpage/site should also show to all users, some example of using multiple repos and how to implement effective includepkgs, exclude, priority etc directives properly for some certain last & STABLE app(s) (which is by default not in CentOS), so that others can understand the pattern, or have a pointer for them. Just mentioning about, that, there is such things called "includepkgs=...", "exclude=..." ad now go do it yourself (and sorry no example), obviously does not help that much to users, and its CentOS's loss as well, users go away to other distros, and ultimately many of them are lost in the jungle. -- Bright Star (Bry8Star).

But you appear to be missing the "C" part in CentOS (or Community Enterprise OS). If you can contribute to the Wiki, then the immediate problem is solved in that such threads can be pointed to the Wiki and slows the growth of my CentOS list folder!

Frankly, if you have a good point to make that would benefit the masses and you have spare time, then it's best to create a Wiki page for it.

Cheers, ak.

On 01/27/2013 07:18 AM, Bry8 Star wrote:

Hi Anthony, it would be really great, to see various types of repo-configs on centos wiki, now if few helpful & experienced users can grab this idea and come forward and share their repo config (and their case/usage scenario along with that), then that would be great.

I'm wondering, why no (experienced) users have already done that yet ! ? (CentOS is not very new at all).

Boxes where i try/test-out new features, apps, etc, I think my repo configs on those will probably be not useful for many & not likened by many, as my need & choices & mods are different than others.

Service providing (production) server box's repo configs are comparatively & usually much simpler.

i will try to share few of my repo-configs in wiki, these are, again, will be based on my own choices+cases+scenario+preferences.

So even if i share my own repo-configs, it will not solve problems of many types of the mass / users.

Someone already EXPERIENCED in CentOS and long time user, should FIRST come forward for the community in WIKI pages, and set an example standard, (amazingly no has yet done that), ... whereas i'm relatively much less-experienced and recently trying to build boxes out of RHEL clone/derivatives (CentOS), with expectation that it might be possible to provide service(s) based on already available various latest released apps, which are STABLE+last+latest.

Every major 3rd party repository out there already provides a Release RPM that sets up their repository. It drops a file in /etc/yum.repos.d/ and the only real thing that one needs to do to those repo files is to set priorities on them if you want to do that.

The problem with 3rd party repos is that they are not necessarily designed to work together. So this means, for example, that EPEL and RPMForge have conflicts between themselves regarding package versions, etc.

Other 3rd party repos might replace packages in the base CentOS distroibutions with older ones.

One thing to remember about Linux is that some packages are built against other packages to share features. As an example, openssh from CentOS-6.3 requires the following packages to build against:

gtk2-devel libX11-devel autoconf automake perl zlib-devel audit-libs-devel >= 2.0.5 util-linux groff man pam-devel tcp_wrappers-devel fipscheck-devel openssl-devel >= 0.9.8j openldap-devel krb5-devel libedit-devel ncurses-devel nss-devel libselinux-devel >= 1.27.7 audit-libs >= 1.0.8 xauth

So, if you upgrade the gtk2 to a new version for some new package, then you would also need to rebuild (or upgrade and rebuild) openssh.

Taking gtk2 as in our example here, if you wanted a new gtk2 in CentOS-5.9, then you would have to rebuild the following packages:

alacarte-0.10.0-1.fc6.src.rpm anaconda-11.1.2.259-1.src.rpm at-spi-1.7.11-3.el5.src.rpm avahi-0.6.16-10.el5_6.src.rpm bluez-gnome-0.5-5.fc6.src.rpm bug-buddy-2.16.0-2.el5.src.rpm control-center-2.16.0-16.el5.src.rpm devhelp-0.12-22.el5_8.src.rpm eclipse-3.2.1-19.el5.centos.src.rpm eel2-2.16.1-1.el5.src.rpm ekiga-2.0.2-7.0.2.src.rpm eog-2.16.0.1-6.el5.src.rpm esc-1.1.0-13.el5_8.2.src.rpm evince-0.6.0-17.el5.src.rpm evolution-2.12.3-19.el5.src.rpm file-roller-2.16.0-2.fc6.src.rpm firefox-10.0.11-1.el5.centos.src.rpm firefox-10.0.12-1.el5.centos.src.rpm gail-1.9.2-3.el5_4.src.rpm gcalctool-5.8.25-1.el5.src.rpm gcc-4.1.2-54.el5.src.rpm GConf2-2.14.0-9.el5.src.rpm gconf-editor-2.16.0-3.el5.src.rpm gdm-2.16.0-59.el5.centos.src.rpm gedit-2.16.0-9.el5.src.rpm gftp-2.0.18-3.2.2.src.rpm ghostscript-8.70-14.el5_8.1.src.rpm gimp-2.2.13-2.0.10.el5.src.rpm glade2-2.12.1-6.el5.src.rpm gnome-applets-2.16.0.1-19.el5.src.rpm gnome-applet-vm-0.1.2-1.el5.src.rpm gnome-bluetooth-0.7.0-10.2.el5.src.rpm gnome-desktop-2.16.0-1.el5.centos.1.src.rpm gnome-games-2.16.0-2.el5.src.rpm gnome-keyring-0.6.0-1.fc6.src.rpm gnome-mag-0.13.1-1.fc6.src.rpm gnome-media-2.16.1-3.el5.src.rpm gnome-mount-0.5-3.el5.src.rpm gnome-netstatus-2.12.0-5.el5.src.rpm gnome-nettool-2.16.0-1.fc6.src.rpm gnome-panel-2.16.1-7.el5.src.rpm gnome-python2-2.16.0-1.fc6.src.rpm gnome-python2-desktop-2.16.0-3.el5.src.rpm gnome-python2-extras-2.14.2-7.el5.src.rpm gnome-screensaver-2.16.1-8.el5_7.5.src.rpm gnome-session-2.16.0-10.el5.src.rpm gnome-system-monitor-2.16.0-4.el5.src.rpm gnome-terminal-2.16.0-5.3.el5_6.1.src.rpm gnome-themes-2.16.0-1.fc6.src.rpm gnome-user-share-0.10-6.el5.src.rpm gnome-utils-2.16.0-5.el5.src.rpm gstreamer-plugins-good-0.10.9-1.el5_3.2.src.rpm gthumb-2.7.8-8.el5.src.rpm gtk2-engines-2.8.0-3.el5.src.rpm gtkhtml2-2.11.0-3.src.rpm gtksourceview-1.8.0-1.fc6.src.rpm gtkspell-2.0.11-2.1.src.rpm gtk-vnc-0.3.8-3.el5.src.rpm gucharmap-1.8.0-1.fc6.src.rpm icon-slicer-0.3-7.2.2.src.rpm im-chooser-0.3.3-6.el5.src.rpm jpilot-0.99.8-7.1.src.rpm kasumi-2.0.1-1.1.fc6.src.rpm libbonoboui-2.16.0-1.fc6.src.rpm libbtctl-0.6.0-9.2.el5.src.rpm libdv-0.104-4.fc6.1.src.rpm libgail-gnome-1.1.3-1.2.1.src.rpm libglade2-2.6.0-2.src.rpm libgnomecanvas-2.14.0-4.1.src.rpm libgnomeprintui22-2.12.1-6.src.rpm libgnomeui-2.16.0-5.el5.src.rpm libgpod-0.4.0-1.el5.src.rpm libgtk-java-2.8.7-3.el5.src.rpm libnotify-0.4.2-6.el5.src.rpm librsvg2-2.16.1-1.el5.src.rpm libwmf-0.2.8.4-10.2.src.rpm libwnck-2.16.0-4.fc6.src.rpm metacity-2.16.0-16.el5.src.rpm mtr-0.71-3.1.src.rpm nautilus-2.16.2-10.el5.src.rpm nautilus-cd-burner-2.16.0-7.el5.src.rpm nautilus-sendto-1.0.1-6.el5.centos.src.rpm NetworkManager-0.7.0-13.el5.src.rpm nmap-4.11-2.src.rpm notification-daemon-0.3.5-9.el5.src.rpm notify-python-0.1.0-3.fc6.src.rpm nspluginwrapper-1.3.0-9.el5.src.rpm openoffice.org-3.1.1-19.10.el5_8.4.src.rpm openssh-4.3p2-82.el5.src.rpm oprofile-0.9.4-20.el5.src.rpm orca-1.0.0-5.el5.src.rpm pidgin-2.6.6-11.el5.4.src.rpm pinentry-0.7.3-3.el5.src.rpm planner-0.14.1-4.el5.src.rpm poppler-0.5.4-19.el5.src.rpm pygtk2-2.10.1-12.el5.src.rpm redhat-artwork-5.1.0-28.el5.centos.src.rpm rhythmbox-0.11.6-4.el5.src.rpm sabayon-2.12.4-9.el5.src.rpm samba3x-3.6.6-0.129.el5.src.rpm sane-frontends-1.0.14-1.2.2.src.rpm scim-1.4.4-44.el5.src.rpm scim-pinyin-0.5.91-16.el5.src.rpm scim-tables-0.5.6-7.src.rpm setools-3.0-3.el5.src.rpm sound-juicer-2.16.0-3.el5.src.rpm thunderbird-10.0.11-1.el5.centos.src.rpm thunderbird-10.0.12-3.el5.centos.src.rpm trousers-0.3.1-4.el5.src.rpm tsclient-0.148-6.el5.src.rpm usermode-1.88-3.el5.2.src.rpm vim-7.0.109-7.2.el5.src.rpm vino-2.13.5-9.el5_4.src.rpm virt-manager-0.6.1-16.el5.src.rpm virt-viewer-0.0.2-3.el5.src.rpm vte-0.14.0-2.el5.src.rpm w3m-0.5.1-18.el5.src.rpm wireshark-1.0.15-5.el5.src.rpm xcdroast-0.98a15-12.2.2.src.rpm xchat-2.6.6-8.el5.src.rpm xen-3.0.3-142.el5.src.rpm xsri-2.1.0-10.fc6.src.rpm xulrunner-10.0.11-1.el5_8.src.rpm xulrunner-10.0.12-1.el5_9.src.rpm yelp-2.16.0-29.el5_8.src.rpm zenity-2.16.0-2.el5.src.rpm

Not only would you need to rebuild all those packages if you upgraded gtk2 ... you might have to find a newer version of the package because the current one might not build against the version of gtk2 you upgrades to.

Also, you would now also need to find and build any packages that build against the packages you just built ... an example here is that a given version of gnome is tied to a certain range gtk2 and if you wanted to upgrade gtk2, you might also have to upgrade gnome ... and that would mean you would also have to upgrade everything that builds against gnome, etc.

So upgrading one package can cause a domino effect that means you have either broken a bunch of packages or you have to rebuild a bunch of packages.

That is why RHEL (and therefore CentOS, as we rebuild their sources) does not do that.

Enterprise Linux is, by definition, designed to maintain the main APIs and ABIs that come out at the beginning of the distribution for the entire lifetime of that main version. So, CentOS-5 will always have Gnome 2.16. Anything that requires a Gnome > 2.16 is likely never going to be released for CentOS-5. The point releases are basically just a point in time "freeze" of the tree to generate install media with updates ... though new functionality is sometimes added and that also takes place during the point releases. But this new functionality is not going to require a new major version of Gnome/GTK or KDE or httpd or openssh or openssl, etc.

Each major version of CentOS is designed to give you its major functionality that it comes with for 10 years. It is not designed to be cutting edge, it is designed to function like it did when you installed it for 10 years from the time it was released. That means if you are an ISP (or Facebook or Zynga) and you invested $1 Million dollars in writing software, you can know how long your software will be able to run before you have to invest to rewrite it for a new platform. You can't rewrite your software every 6 months when the latest version of Gnome is released. You need a solid, stable platform to run your code for a long time. That is Enterprise Linux.

Some of the Main versions (CentOS-5 and CentOS-6 and CentOS-7) will be active at the same time, and you can pick the main tree that provides the version that you need and you can then stabilize on it. Right now, both CentOS-5 (active until 2017) and CentOS-6 (active until 2020) are both active. They will both be secure until their EOL dates by a process that Red Hat calls backporting:

https://access.redhat.com/security/updates/backporting/

If one is looking for the latest and greatest Linux features on the cutting edge, well CentOS is not for that. Fedora would be the Red Hat product that is cutting edge ... or Rawhide if you are looking for bleeding edge.

Debian has the same sort of thing. They have oldstable (gets security updates for 1 year after a stable release), stable (the current release), testing (the next release), and unstable (cutting edge, named sid).

Rawhide is like sid, Fedora is like the Testing release, RHEL is like the Stable Release and the Old Stable release. Except RHEL (and CentOS since we build those sources) provides those stable releases for a lot longer period of time.

Lenny is Debain's Old Stable and it no longer gets security updates ... but it was active from February 2009 to February 2012. That is only 3 years. CentOS is around for 10 years.

With CentOS, you normally get to pick when you need to upgrade based on your requirements, not an artificial time period of 3 years maximum. However, you can, if you want upgrade to a new version of CentOS every 2-3 years as well ... it is just not required to get security updates.

For example, CentOS-4 was released in March 2005 and CentOS-5 was released in April 2007 CentOS-6 was released in July 2010. So, the ability to upgrade (if you want) every couple of years is there as well as going up to 10 years between upgrades (if you choose to).

My point is that if you are expecting CentOS to be Debian from SID to OldStable or OpenSUSE or Fedora ... it is not that, nor is it meant to be that. It is meant to be a long term stable release.

For example, IF CentOS/RHEL (based/clone/derivative linux) has a specific app/lib at v7.00, and if the source/origin/upstream developer has released a STABLE version v9.00 (with new features and older bugs fixed), and when more bugs are fixed later on for the v9.00, then how that patch/code is re-implemented on v7.00 ! ? would not a v9.00 bug-fix/patch cause more problem(s) when applied on v7.00 ?

or, is it this case, that, when a bug-fix/patch is applied on an older v7.00 app/lib by RHEL upstream developers (or by origin/source upstream developers), and then, that is, later re-applied/re-implemented on RHEL/CentOS (clone/derivative) v7.00 app/lib ?

So, does that mean that a CentOS/RHEL based clone/derivative linux's v7.00 app/lib, will always lack new v9.00 features ? until, few months/yrs later when CentOS/RHEL clone/derivative also gets finally updated to v9.00 ? but by then source/origin upstream developers most likely already released v11.00 or so ? or, is it this case, that, when some features of v9.00 are transferred into v7.0, then that is also re-implemented by CentOS/RHEL ? is this really happens ? or, is it, that, CentOS/RHEL based linux developers decide to add v8.00 in CentOS/RHEL, as an update to the previous v7.00, and then that is re-applied over the previous v7.00 ? but is not by then RHEL or source/origin upstream/developers already released v10.00, after v9.00

Some items on the Desktop (like Firefox, Thunderbird, LibreOffice, etc.) are upgraded inside a major release, but most things are not going to be. If CentOS-5.0 ships with Gnome 2.16 ... Gnome 2.16 is going to be the version it ends with as well in 10 years (it will get security updates and may go from 2.16.1 to 2.16.8 or whatever, but it will still be 2.16.x at the end).

CentOS is not designed to major upgrades in place. Because CentOS does not change major ABIs and APIs inside a major release, you can use the normal tools (like yum) to upgrade between the point releases in a given major release (that would be CentOS-5.x to any other CentOS-5.x, ie, upgrade from 5.2 to 5.9 is supported). However the ABIs and APIs can have major changes between major releases and therefore your configuration files and other things (like data, etc) will likely require changes if you move to a new major version (that would be moving from a CentOS-5.x release to a CentOS-6.x release).

CentOS is what it is, so please make sure you are using it for what it is meant for, not trying to make it be one of the cutting edge, release every 6 months, distros of Lunix ... if that is the kind of Linux you want then CentOS is not for you.

Thank you very much, Johnny Hughes, Nikolaos Milas, (and others), for nice explanation & example.

If ALL new apps/libs starts to change their API without backward compatibility, then, definitely updating/upgrading core/base apps would cause domino effect, like you have pointed out. Various apps and libs are inter-dependent & inter-connected with each others.

But do all ALL apps, or, libs do really change ALL of their API set ? My understanding is, some new APIs are added for new features, and older same API gets more refined, and/or newer query or response parameters are added behind the old existing query/response.

Do not developers (when coding for newer version), anticipate ? ... that their app/lib will and may try to communicate with another older or another advanced (with extra query fields) app/lib, so based on that it need to provide answer (by adding extra new response/parameters after the older API standard/parameters) or take action ?!

I'm sure smart developers do really able to and do really code, so that their app/lib does not fail to receive & answer when interacting with an older app/lib or a newer app/lib.

If they(developers) purposefully does not honor backward compatibility handling in their app/lib, that is kind of like, forcing people to buy new computer/hardware for new Windows8.

Why apps/libs are getting bigger in src or in their binary size ? with what ! ?

-- Bright Star.

Received from Nikolaos Milas, on 2013-01-27 4:56 PM:

On 27/1/2013 5:11 μμ, Johnny Hughes wrote:

...

So upgrading one package can cause a domino effect that means you have either broken a bunch of packages or you have to rebuild a bunch of packages.

That is why you should *only* upgrade what is VITAL for your application(s), and in a carefully planned manner, as Johnny explains.

Building on the example I gave earlier, OpenLDAP Server, which IS important to be current in production due to critical bug (not only security) fixes, can be safely upgraded using LTB project RPMs (http://ltb-project.org/wiki/) which allows a current version WITHOUT affecting the rest of the system (by using alternate paths to install libs, executables etc.).

Any other Servers in need for a current version can be safely upgraded through a careful approach to make sure they don't break other things.

So, get to know your OS, your apps, your (S)RPMs; then plan, test on a non-production system, then plan deployment in production. For example, we *only* upgrade other critical apps like Postfix, Dovecot etc. when running as enterprise servers. There is no other option here; for example, Postfix as available in RHEL/CentOS 5, is no more supported by the Postfix developer. We have to build our own RPMs based on our particular requirements.

We enjoy the stability of CentOS because it allows us to only upgrade what is vital. We don't care about Gnome and other modernization(s). Our CentOS 5 enterprise servers do not even have Gnome installed.

Nick _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

On 27/1/2013 5:11 μμ, Johnny Hughes wrote:

Taking gtk2 as in our example here, if you wanted a new gtk2 in CentOS-5.9, then you would have to rebuild the following packages:

A quick (?) question:

What is the best way to view dependencies?

1. We can use yum deplist <xxx.rpm> to view dependencies of a particular installed package, but we would like to limit output to only *already installed* packages and not view all suitable ones. Also, we would like to view dependencies as a full rpm name and not as they are output by default (e.g. not "glibc.x86_64 2.12-1.80.el6_3.6" but "glibc-2.12-1.80.el6_3.6.x86_64")

2. Is there a way to find out the opposite, i.e. which *installed* packages are dependent on a particular installed package? In your example above (about gtk2), how could we have produced the list of dependent-on-it packages which you have kindly provided?

Thanks, Nick

I've tried this command:

yum deplist gnutls --disableplugin=filter-data,priorities --disableexcludes=all --enablerepo=* > yum_01.txt

... that yum_01.txt file was around 202KB in my case.

Result is formatted like this: ##<...snip...> package: gnutls.i686 2.12.22-1.fc18 dependency: libz.so.1 provider: zlib.i686 1.2.3-27.el6 provider: zlib.i686 1.2.7-9.fc18 provider: zlib.i686 1.2.3-25.el6 provider: zlib.i686 1.2.3-27.el6 provider: zlib.i686 1.2.3-27.el6 provider: zlib.i686 1.2.3-25.el6 provider: zlib.i686 1.2.3-27.el6 provider: zlib.i686 1.2.3-25.el6 provider: zlib.i686 1.2.3-27.el6 dependency: libgcrypt.so.11 provider: libgcrypt.i686 1.5.0-8.fc18 provider: libgcrypt.i686 1.4.5-9.el6_2.2 provider: libgcrypt.i686 1.5.0-6.fc18 provider: libgcrypt11.i686 1.4.0-15.el6 provider: libgcrypt.i686 1.4.5-9.el6 provider: libgcrypt.i686 1.4.5-5.el6 provider: libgcrypt.i686 1.4.5-3.el6 provider: libgcrypt.i686 1.4.5-5.el6 provider: libgcrypt.i686 1.4.5-5.el6_1.2 provider: libgcrypt.i686 1.4.5-9.el6 provider: libgcrypt.i686 1.4.5-9.el6_2.2 provider: libgcrypt.i686 1.4.5-9.el6_2.2 provider: libgcrypt.i686 1.4.5-3.el6 provider: libgcrypt.i686 1.4.5-9.el6_2.2 dependency: libgcrypt >= 1.2.2 provider: libgcrypt.i686 1.5.0-8.fc18 provider: libgcrypt.i686 1.4.5-9.el6_2.2 provider: libgcrypt.i686 1.5.0-6.fc18 provider: libgcrypt.i686 1.4.0-15.el6 provider: libgcrypt.i686 1.4.5-9.el6 ##<...snip...>

the problem is ... in the "provider" lines, yum is not showing repo name, after showing lib/pkg name & version, on each line !

If repo, pkg/lib, version, arch, etc are shown & known ... Then it would have been bit easier to build a technique what to keep and what to update.

-- Bright Star.

Received from Johnny Hughes,, on 2013-01-28 8:17 PM:

On 01/28/2013 01:26 PM, Nikolaos Milas wrote:

...

On 27/1/2013 5:11 μμ, Johnny Hughes wrote:

...

Taking gtk2 as in our example here, if you wanted a new gtk2 in CentOS-5.9, then you would have to rebuild the following packages:

A quick (?) question:

What is the best way to view dependencies?

1. We can use yum deplist <xxx.rpm> to view dependencies of a

particular installed package, but we would like to limit output to only *already installed* packages and not view all suitable ones. Also, we would like to view dependencies as a full rpm name and not as they are output by default (e.g. not "glibc.x86_64 2.12-1.80.el6_3.6" but "glibc-2.12-1.80.el6_3.6.x86_64")

2. Is there a way to find out the opposite, i.e. which

*installed* packages are dependent on a particular installed package? In your example above (about gtk2), how could we have produced the list of dependent-on-it packages which you have kindly provided?

With respect to build and SRPMS, I do this:

I go to a place where all the distro's SRPMS (we have that place on vault), then I do this command:

rpm -qp --requires <name>.src.rpm

That will tell me everything that is required to build that individual SRPM.

However, you would also need to take every requirement and do the same thing to it ... in other words, find the requirements of each requirement.

There is no easy way to do this (that I know of) ... if I want to know, I do the work.

Also, if you want to develop an initial list of SRPMS that are impacted when changing something, you would do this inside the "LATEST" SRPMS (again, using gtk2 as the example):

for srpm in $(ls *.src.rpm); do requires_gtk2=$(rpm -qp --requires $srpm | grep "gtk2-devel"); if [ "$requires_gtk2" != "" ]; then echo $srpm; fi; done

(the above is one line if wrapped)

Once you develop the list of SRPMS, you would need to develop another list of requirements for each requirement.

There is a yum-utility called yum-builddep that will install all the dependencies to build a package for you ... you can do that into a minimal installed VM as well.

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

On 01/28/2013 04:43 AM, Mark LaPierre wrote:

On 01/27/2013 08:18 AM, Bry8 Star wrote:

...

Hi Anthony, it would be really great, to see various types of repo-configs on centos wiki, now if few helpful& experienced users can grab this idea and come forward and share their repo config (and their case/usage scenario along with that), then that would be great.

<snip>

(Sorry for spelling& grammar mistakes in previous and in this posting, pls kindly disregard, its not a grammar discussion thread).

-- Bry8Star.

Received from Anthony K, on 2013-01-27 2:48 AM:

...

On 26/01/13 14:59, Bry8 Star wrote:

...

CentOS webpage/site should also show to all users, some example of using multiple repos and how to implement effective includepkgs, exclude, priority etc directives properly for some certain last& STABLE app(s) (which is by default not in CentOS), so that others can understand the pattern, or have a pointer for them. Just mentioning about, that, there is such things called "includepkgs=...", "exclude=..." ad now go do it yourself (and sorry no example), obviously does not help that much to users, and its CentOS's loss as well, users go away to other distros, and ultimately many of them are lost in the jungle. -- Bright Star (Bry8Star).

But you appear to be missing the "C" part in CentOS (or Community Enterprise OS). If you can contribute to the Wiki, then the immediate problem is solved in that such threads can be pointed to the Wiki and slows the growth of my CentOS list folder!

Frankly, if you have a good point to make that would benefit the masses and you have spare time, then it's best to create a Wiki page for it.

Cheers, ak.

There is already a fine page on this subject on the wiki.

http://wiki.centos.org/AdditionalResources/Repositories

The point being made is that various people have the knowledge and experience to advise a startup setting for priority= for each repo I know that what I'm using has caused conflicts that have been quite time consuming to resolve - what works for others would be most helpful I do recognise that this will vary depending upon what tools are required but as a start: 1. developer workstation - what repos and what priority 2. LAMP server - probably just CentOS repos and something which deals with later php / perl / ruby 3. web / internet workstation - needs audio and video stuff working just my thoughts for starters.

On 01/28/2013 01:50 PM, Johnny Hughes wrote:

On 01/27/2013 06:20 PM, Rob Kampen wrote:

...

On 01/28/2013 04:43 AM, Mark LaPierre wrote:

...

On 01/27/2013 08:18 AM, Bry8 Star wrote:

...

Hi Anthony, it would be really great, to see various types of repo-configs on centos wiki, now if few helpful& experienced users can grab this idea and come forward and share their repo config (and their case/usage scenario along with that), then that would be great.

<snip>

(Sorry for spelling& grammar mistakes in previous and in this posting, pls kindly disregard, its not a grammar discussion thread).

-- Bry8Star.

Received from Anthony K, on 2013-01-27 2:48 AM:

...

On 26/01/13 14:59, Bry8 Star wrote:

...

CentOS webpage/site should also show to all users, some example of using multiple repos and how to implement effective includepkgs, exclude, priority etc directives properly for some certain last& STABLE app(s) (which is by default not in CentOS), so that others can understand the pattern, or have a pointer for them. Just mentioning about, that, there is such things called "includepkgs=...", "exclude=..." ad now go do it yourself (and sorry no example), obviously does not help that much to users, and its CentOS's loss as well, users go away to other distros, and ultimately many of them are lost in the jungle. -- Bright Star (Bry8Star).

But you appear to be missing the "C" part in CentOS (or Community Enterprise OS). If you can contribute to the Wiki, then the immediate problem is solved in that such threads can be pointed to the Wiki and slows the growth of my CentOS list folder!

Frankly, if you have a good point to make that would benefit the masses and you have spare time, then it's best to create a Wiki page for it.

Cheers, ak.

There is already a fine page on this subject on the wiki.

http://wiki.centos.org/AdditionalResources/Repositories

The point being made is that various people have the knowledge and experience to advise a startup setting for priority= for each repo I know that what I'm using has caused conflicts that have been quite time consuming to resolve - what works for others would be most helpful I do recognise that this will vary depending upon what tools are required but as a start:

1. developer workstation - what repos and what priority
2. LAMP server - probably just CentOS repos and something which deals

with later php / perl / ruby 3. web / internet workstation - needs audio and video stuff working just my thoughts for starters.

That totally depends on what you need to install and what repo it is in. Since 3rd party repos are constantly adding new packages that they did not have last week, it is impossible to say what would be the proper priorities.

I already posted what I personally do, which is:

Install CentOS and set Base, updates, extras, and fasttrack to a Priority=1

I usually do not need to enable centosplus, but if I do, I set it to Priority=2 and I put "excludes=<pkg_names>" in the Priority=1 repos for the packages I want let CentOS plus replace in those repos.

I then normally add EPEL and set the Priority=10 for that.

Hopefully, that is all I need to add.

If I have to add any more repositories, first make sure my packages are currently all updated by doing a yum upgrade. Then I add the new repos one at a time and I make them Priority=10 (the same as EPEL) ... and after I add them , i do a "yum update". If it tries to update, I look at the packages and decide if I am going to allow the update or not ... if I am ok to do the updates, then I do them and make sure it works. Then I would install the packages I need from that repo. Then I would add the next new repo till I get to the end.

The best scenario is that all your 3rd party repos can co-exist at the same Priority setting and that is where I start (at Priority=10) ... and if something does not work, I troubleshoot it and take individual action.

Each individual machine is going to require a unique and separate group of settings based on what you want to install ... which is why there is no official recommendations.

I personally am using the following repos right now on my main Desktop, which is CentOS-6.3:

adobe-linux-x86_64 | 951 B 00:00 base | 3.7 kB 00:00 cr | 3.0 kB 00:00 elrepo | 1.9 kB 00:00 elrepo-extras | 1.9 kB 00:00 extras | 3.5 kB 00:00 fasttrack | 3.5 kB 00:00 google-chrome | 951 B 00:00 google-musicmanager | 951 B 00:00 livna | 1.3 kB 00:00 nux-dextop | 2.7 kB 00:00 rpmforge | 1.9 kB 00:00 updates | 3.5 kB 00:00

All of the secondary repos are set to the same priorities and everything seems to work.

Right now I have an "exclude=wxGtk*" for rpmforge for some reason. And an "exclude=nx freenx*" for Nux! repo.

thanks Johnny - that was helpful (for me anyhow).

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

Sorry, mentioned [pgdg92] twice. Pls ignore/delete the [pgdg92] that has priority 15.

Received from Bry8 Star, on 2013-01-28 10:16 AM:

For PostgreSQL, i've done these (shown below) at initial/test stage: (pls DO NOT follow/copy it, try to understand pattern and do what fits for your case/need).

From a VE instance inside HN:

yum has these plugins: fastestmirror, filter-data, keys, list-data, merge-conf, presto, priorities, security, verify.

# CentOS-Base.repo [CentOS-base] #<...snip...> enabled=1 # i do not want below set from centos repo # as these exacts are released by postgre devs # but other postgre portions from others i do want exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOS-updates] #<...snip...> enabled=1 exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOS-extras] #<...snip...> enabled=1 exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOSplus] #<...snip...> enabled=1 exclude=kernel.* kernel-firmware.* kernel-headers.* postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

# pgdg92-centos.repo [pgdg92] #<...snip...> enabled=1 priority=10

[pgdg92-source] #<...snip...> enabled=1 priority=10

#scntflnx.repo [SciLnx-6x-os] #<...snip...> enabled=1 exclude=redhat-rpm-config.* postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-updates-fastbugs] #<...snip...> enabled=1 exclude=postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-updates-security] #<...snip...> enabled=1 exclude=postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-debuginfo] #<...snip...> enabled=0 priority=13

[SciLnx-6x-SRPMS] #<...snip...> enabled=1 priority=13

[pgdg92] #<...snip...> enabled=1 priority=15

... other repo & distro ...

In above way, i get most from PosteGre repo, and other postgre related side packages from other repos. if i were to place exclude=postgre* then all packages would have gotten excluded, and then the extra postgre pkg which are not released by postgre, would get hidden/excluded as well, though, i could have used this approach for to get that:

includepkgs=related-extra-postgre-pkg

to get those. but then i would need specific name, but what if new extra pkg that i dont know yet, is included, but since i have not specified specifically, those would get hidden/excluded. (my repo inquiries have shown, many other repo has at-least 1 to 3 or some up to 6 extra postgre related tools which are not released by postgre devs themselves).

Few patterns/logics which i like to follow:

repo#1, rpm priority 1, srpm priority 50 repo#2, rpm priority 10, srpm priority 40 repo#3, rpm priority 15, srpm priority 30 repo#4, rpm priority 20, srpm priority 20 ...

Priority numbering pattern/logic: Since i'm using centos as my base, so various DISTRO & REPO rpm/binary which are closest to CentOS & CentOS's source RHEL, those distro & repo will get lowest priority# close to centos. i like to think/visualize in this way ... a LIST from top to down, now number them from top to bottom 1, 2, 3.. CentOS is at position 1, the top-most position. The lowest priority number, is at highest, upper position in the list, is chosen first by yum/installer, and has highest priority, if it(yum) wants to install something.

my knol is this (i could be wrong): RHEL based source -> |-> CentOS base. |-> Scientific Linux. |-> Oracle Linux. |-> ClearOS Linux.

(Fedora stable source -> RHEL -> RHEL source).

EPEL has extra small tools/apps/libs for RHEL base apps/libs, and also for RHEL based clone/derivative distros (few mentioned above). RPMforge/dag has many perl related stuff. REMI has latest+stable PHP, MySql related stuff. ELrepo is for hardware latest+stable drivers, kernels. kdeRedhat is mostly for latest+stable KDE, QT, samba, etc. RPMFusion has closed-source-free, open-source-free, non-free etc apps. ATrpms has MythTV, Scientific Apps, etc. LnxTechNet has audio,video,etc apps.

And Fedora 17, 18 source/src/SRPMS are needed to be re-built on cnetos for those apps/libs, then those can be used with centos.

Pkg "Excluding" pattern/logic: (means, Getting specific pkg from Non-Centos repo): Now for example if i want something specific called "pkg-set-3" from repo#3, then i would have to place those specific pkgs inside the "exclude" line in repo(s) which are listed at upper position on the repo-list, so that will be repo#2, repo#1, so it will be: # repo_1.repo file inside /etc/yum.repos.d [repo_1] exclude=pkg-set-3

# repo_2.repo file inside /etc/yum.repos.d [repo_2] exclude=pkg-set-3 # in 'exclude' pkgs are 'space' separated

Where & How to tweak & apply intelligence/knowledge and experience and tricks , etc for repo-config ? i like to call it HI (Human Intelligence), multiple (or one) developer's initial HI becomes AI in a software.

Each repo file will most likely have one or more section/channel like below: [section-or-channel-name] config_options=values ...

in repo#2, the "pkg-set-3" can be placed inside all section/channel of that repo file.

For setting repo#1 (in our case, this is centos), we will have to do these:

set "enabled=" to 0 in all new repo, (except centos.repo), or, by using yum command-line options instruct "yum" to disable all repo, then instruct yum to enable only centos-base,centos-updates, etc repo only and then do "yum check-update", like this:

yum --disablerepo=* --enablerepo=base,update,centosplus,extra check-update

... above command will show if centos repo has new updates (or no updates) for your system.

then set enabled=1 in those section/channel which you need for the app/lib (pk-set-3) which you wanted.

then run do "yum check-update". ... It will show list of app/lib which will be updated if you were to run "yum update" or "yum upgrade".

Also see other helpful yum commands.

Since these yum commands will show app/lib name and which repo has it, observe carefully and make a hand-written list first (or write on a text file on the client computer from where you're connecting to your centos server which you are trying to configure), a list, apps/libs which you need to exclude from which repo.

Other helpful YUM COMMANDS:

yum list updates '*' --disableplugin=filter-data,priorities --disableexcludes=all

... should show what updates will be in queue if all "excludes" are ignored.

yum list all available 'pkg-name*' --disableplugin=filter-data,priorities --disableexcludes=all

... should show which of your repos has any package that are close to the name "pkg-name" and will also show their version # and what repo has it.

Anytime you change any "somename.repo" file, then its better to do first: yum check-update (but do not update, unless you are sure which app is coming from which exact repo and if that is what you want or not).

Please ADD/POST MORE/YOUR HELPFUL COMMANDS, and tell us what it does, where useful.

Since source/SRPMS is/are needed when you/i need to make src.rpm, so distro & repo closest to CentOS gets higher priority, usually that is in reverse order than the RPM/binaries.

Please correct my mistakes and add your responses, ideas, suggestions, logics, patterns, etc.

Thanks in advance,

-- Bright Star.

Received from Johnny Hughes,, on 2013-01-28 12:50 AM:

...

On 01/27/2013 06:20 PM, Rob Kampen wrote:

...

On 01/28/2013 04:43 AM, Mark LaPierre wrote:

...

On 01/27/2013 08:18 AM, Bry8 Star wrote:

...

Hi Anthony, it would be really great, to see various types of repo-configs on centos wiki, now if few helpful& experienced users can grab this idea and come forward and share their repo config (and their case/usage scenario along with that), then that would be great.

<snip>

(Sorry for spelling& grammar mistakes in previous and in this posting, pls kindly disregard, its not a grammar discussion thread).

-- Bry8Star.

Received from Anthony K, on 2013-01-27 2:48 AM:

...

On 26/01/13 14:59, Bry8 Star wrote: > CentOS webpage/site should also show to all users, > some example of using multiple repos and how to > implement effective includepkgs, exclude, priority > etc directives properly for some certain last& > STABLE app(s) (which is by default not in CentOS), so > that others can understand the pattern, or have a > pointer for them. Just mentioning about, that, there > is such things called "includepkgs=...", > "exclude=..." ad now go do it yourself (and sorry no > example), obviously does not help that much to users, > and its CentOS's loss as well, users go away to other > distros, and ultimately many of them are lost in the > jungle. -- Bright Star (Bry8Star). But you appear to be missing the "C" part in CentOS (or Community Enterprise OS). If you can contribute to the Wiki, then the immediate problem is solved in that such threads can be pointed to the Wiki and slows the growth of my CentOS list folder!

Frankly, if you have a good point to make that would benefit the masses and you have spare time, then it's best to create a Wiki page for it.

Cheers, ak.

There is already a fine page on this subject on the wiki.

http://wiki.centos.org/AdditionalResources/Repositories

The point being made is that various people have the knowledge and experience to advise a startup setting for priority= for each repo I know that what I'm using has caused conflicts that have been quite time consuming to resolve - what works for others would be most helpful I do recognise that this will vary depending upon what tools are required but as a start: 1. developer workstation - what repos and what priority 2. LAMP server - probably just CentOS repos and something which deals with later php / perl / ruby 3. web / internet workstation - needs audio and video stuff working just my thoughts for starters.

That totally depends on what you need to install and what repo it is in. Since 3rd party repos are constantly adding new packages that they did not have last week, it is impossible to say what would be the proper priorities.

I already posted what I personally do, which is:

Install CentOS and set Base, updates, extras, and fasttrack to a Priority=1

I usually do not need to enable centosplus, but if I do, I set it to Priority=2 and I put "excludes=<pkg_names>" in the Priority=1 repos for the packages I want let CentOS plus replace in those repos.

I then normally add EPEL and set the Priority=10 for that.

Hopefully, that is all I need to add.

If I have to add any more repositories, first make sure my packages are currently all updated by doing a yum upgrade. Then I add the new repos one at a time and I make them Priority=10 (the same as EPEL) ... and after I add them , i do a "yum update". If it tries to update, I look at the packages and decide if I am going to allow the update or not ... if I am ok to do the updates, then I do them and make sure it works. Then I would install the packages I need from that repo. Then I would add the next new repo till I get to the end.

The best scenario is that all your 3rd party repos can co-exist at the same Priority setting and that is where I start (at Priority=10) ... and if something does not work, I troubleshoot it and take individual action.

Each individual machine is going to require a unique and separate group of settings based on what you want to install ... which is why there is no official recommendations.

I personally am using the following repos right now on my main Desktop, which is CentOS-6.3:

adobe-linux-x86_64 | 951 B 00:00 base | 3.7 kB 00:00 cr | 3.0 kB 00:00 elrepo | 1.9 kB 00:00 elrepo-extras | 1.9 kB 00:00 extras | 3.5 kB 00:00 fasttrack | 3.5 kB 00:00 google-chrome | 951 B 00:00 google-musicmanager | 951 B 00:00 livna | 1.3 kB 00:00 nux-dextop | 2.7 kB 00:00 rpmforge | 1.9 kB 00:00 updates | 3.5 kB 00:00

All of the secondary repos are set to the same priorities and everything seems to work.

Right now I have an "exclude=wxGtk*" for rpmforge for some reason. And an "exclude=nx freenx*" for Nux! repo.

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

When using multiple repo, then how would i know, an app/lib from non-CentOS (that is, 3rd party repo) is going to conflict or not with CentOS core, base, essential app/lib packages ?

We need to know CentOS core, base, essential components.

According to this file:

/usr/lib/anaconda/installclasses/rhel.py

That is located inside the mounted image of this below file: http://mirror.centos.org/centos-6/6/os/x86_64/images/install.img

or http://mirror.centos.org/centos-6/6/os/x86/images/install.img

* A CentOS Basic Server has : base, console-internet, core, debugging, directory-client, hardware-monitoring, java-platform, large-systems, network-file-system-client, performance, perl-runtime, server-platform. * Database Server : Basic Server (shown above), and, mysql-client, mysql, postgresql-client, postgresql, system-admin-tools. * Web Server: Basic Server (shown above), and, mysql-client, php, postgresql-client, turbogears, web-server, web-servlet.

So below command will show what app/lib/pkg are in "Basic Server", and should not be replaced by a 3rd party repo packages:

yum groupinfo base core "Console internet tools" "Directory Client" "Java Platform" "Network file system client" "Server Platform" debugging large-systems performance perl-runtime hardware-monitoring --disablerepo=* --enablerepo=base,extra,update --disableplugin=filter-data,priorities --disableexcludes=all

in above, it is, one command-line, (text-based email-system will break it, so remove LF/CR char to get full command-line )

You must try to avoid replacing those, (centos essential packages).

Copy the list in a text file, remove lines(LF/CR), make paragraphs, and print it out or keep the text file easily accessible location. So you can see it to decide, on 3rd party repo which package should be placed inside "exclude".

If there is/are mistake(s) in above suggestion, please correct. Please post better, other way(s) to achieve these.

Thanks, -- Bright Star.

Received from Bry8 Star, on 2013-01-28 10:16 AM:

For PostgreSQL, i've done these (shown below) at initial/test stage: (pls DO NOT follow/copy it, try to understand pattern and do what fits for your case/need).

From a VE instance inside HN:

yum has these plugins: fastestmirror, filter-data, keys, list-data, merge-conf, presto, priorities, security, verify.

# CentOS-Base.repo [CentOS-base] #<...snip...> enabled=1 # i do not want below set from centos repo # as these exacts are released by postgre devs # but other postgre portions from others i do want exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOS-updates] #<...snip...> enabled=1 exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOS-extras] #<...snip...> enabled=1 exclude=postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

[CentOSplus] #<...snip...> enabled=1 exclude=kernel.* kernel-firmware.* kernel-headers.* postgresql.* postgresql-contrib* postgresql-debuginfo* postgresql-devel* postgresql-docs* postgresql-jdbc.* postgresql-jdbc-debuginfo* postgresql-libs* postgresql-odbc.i686 postgresql-odbc-debuginfo* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-tcl.i686 postgresql-tcl-debuginfo* postgresql-test* priority=1

# pgdg92-centos.repo [pgdg92] #<...snip...> enabled=1 priority=10

[pgdg92-source] #<...snip...> enabled=1 priority=10

#scntflnx.repo [SciLnx-6x-os] #<...snip...> enabled=1 exclude=redhat-rpm-config.* postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-updates-fastbugs] #<...snip...> enabled=1 exclude=postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-updates-security] #<...snip...> enabled=1 exclude=postgresql.* postgresql-devel* postgresql-docs* postgresql-contrib* postgresql-jdbc.* postgresql-libs* postgresql-odbc* postgresql-plperl* postgresql-plpython* postgresql-pltcl* postgresql-server* postgresql-test* priority=13

[SciLnx-6x-debuginfo] #<...snip...> enabled=0 priority=13

[SciLnx-6x-SRPMS] #<...snip...> enabled=1 priority=13

[pgdg92] #<...snip...> enabled=1 priority=15

... other repo & distro ...

In above way, i get most from PosteGre repo, and other postgre related side packages from other repos. if i were to place exclude=postgre* then all packages would have gotten excluded, and then the extra postgre pkg which are not released by postgre, would get hidden/excluded as well, though, i could have used this approach for to get that:

includepkgs=related-extra-postgre-pkg

to get those. but then i would need specific name, but what if new extra pkg that i dont know yet, is included, but since i have not specified specifically, those would get hidden/excluded. (my repo inquiries have shown, many other repo has at-least 1 to 3 or some up to 6 extra postgre related tools which are not released by postgre devs themselves).

Few patterns/logics which i like to follow:

repo#1, rpm priority 1, srpm priority 50 repo#2, rpm priority 10, srpm priority 40 repo#3, rpm priority 15, srpm priority 30 repo#4, rpm priority 20, srpm priority 20 ...

Priority numbering pattern/logic: Since i'm using centos as my base, so various DISTRO & REPO rpm/binary which are closest to CentOS & CentOS's source RHEL, those distro & repo will get lowest priority# close to centos. i like to think/visualize in this way ... a LIST from top to down, now number them from top to bottom 1, 2, 3.. CentOS is at position 1, the top-most position. The lowest priority number, is at highest, upper position in the list, is chosen first by yum/installer, and has highest priority, if it(yum) wants to install something.

my knol is this (i could be wrong): RHEL based source -> |-> CentOS base. |-> Scientific Linux. |-> Oracle Linux. |-> ClearOS Linux.

(Fedora stable source -> RHEL -> RHEL source).

EPEL has extra small tools/apps/libs for RHEL base apps/libs, and also for RHEL based clone/derivative distros (few mentioned above). RPMforge/dag has many perl related stuff. REMI has latest+stable PHP, MySql related stuff. ELrepo is for hardware latest+stable drivers, kernels. kdeRedhat is mostly for latest+stable KDE, QT, samba, etc. RPMFusion has closed-source-free, open-source-free, non-free etc apps. ATrpms has MythTV, Scientific Apps, etc. LnxTechNet has audio,video,etc apps.

And Fedora 17, 18 source/src/SRPMS are needed to be re-built on cnetos for those apps/libs, then those can be used with centos.

Pkg "Excluding" pattern/logic: (means, Getting specific pkg from Non-Centos repo): Now for example if i want something specific called "pkg-set-3" from repo#3, then i would have to place those specific pkgs inside the "exclude" line in repo(s) which are listed at upper position on the repo-list, so that will be repo#2, repo#1, so it will be: # repo_1.repo file inside /etc/yum.repos.d [repo_1] exclude=pkg-set-3

# repo_2.repo file inside /etc/yum.repos.d [repo_2] exclude=pkg-set-3 # in 'exclude' pkgs are 'space' separated

Where & How to tweak & apply intelligence/knowledge and experience and tricks , etc for repo-config ? i like to call it HI (Human Intelligence), multiple (or one) developer's initial HI becomes AI in a software.

Each repo file will most likely have one or more section/channel like below: [section-or-channel-name] config_options=values ...

in repo#2, the "pkg-set-3" can be placed inside all section/channel of that repo file.

For setting repo#1 (in our case, this is centos), we will have to do these:

set "enabled=" to 0 in all new repo, (except centos.repo), or, by using yum command-line options instruct "yum" to disable all repo, then instruct yum to enable only centos-base,centos-updates, etc repo only and then do "yum check-update", like this:

yum --disablerepo=* --enablerepo=base,update,centosplus,extra check-update

... above command will show if centos repo has new updates (or no updates) for your system.

then set enabled=1 in those section/channel which you need for the app/lib (pk-set-3) which you wanted.

then run do "yum check-update". ... It will show list of app/lib which will be updated if you were to run "yum update" or "yum upgrade".

Also see other helpful yum commands.

Since these yum commands will show app/lib name and which repo has it, observe carefully and make a hand-written list first (or write on a text file on the client computer from where you're connecting to your centos server which you are trying to configure), a list, apps/libs which you need to exclude from which repo.

Other helpful YUM COMMANDS:

yum list updates '*' --disableplugin=filter-data,priorities --disableexcludes=all

... should show what updates will be in queue if all "excludes" are ignored.

yum list all available 'pkg-name*' --disableplugin=filter-data,priorities --disableexcludes=all

... should show which of your repos has any package that are close to the name "pkg-name" and will also show their version # and what repo has it.

Anytime you change any "somename.repo" file, then its better to do first: yum check-update (but do not update, unless you are sure which app is coming from which exact repo and if that is what you want or not).

Please ADD/POST MORE/YOUR HELPFUL COMMANDS, and tell us what it does, where useful.

Since source/SRPMS is/are needed when you/i need to make src.rpm, so distro & repo closest to CentOS gets higher priority, usually that is in reverse order than the RPM/binaries.

Please correct my mistakes and add your responses, ideas, suggestions, logics, patterns, etc.

Thanks in advance,

-- Bright Star.

Received from Johnny Hughes,, on 2013-01-28 12:50 AM:

...

On 01/27/2013 06:20 PM, Rob Kampen wrote:

...

On 01/28/2013 04:43 AM, Mark LaPierre wrote:

...

On 01/27/2013 08:18 AM, Bry8 Star wrote:

...

Hi Anthony, it would be really great, to see various types of repo-configs on centos wiki, now if few helpful& experienced users can grab this idea and come forward and share their repo config (and their case/usage scenario along with that), then that would be great.

<snip>

(Sorry for spelling& grammar mistakes in previous and in this posting, pls kindly disregard, its not a grammar discussion thread).

-- Bry8Star.

Received from Anthony K, on 2013-01-27 2:48 AM:

...

On 26/01/13 14:59, Bry8 Star wrote: > CentOS webpage/site should also show to all users, some > example of using multiple repos and how to implement > effective includepkgs, exclude, priority etc directives > properly for some certain last& STABLE app(s) (which > is by default not in CentOS), so that others can > understand the pattern, or have a pointer for them. > Just mentioning about, that, there is such things > called "includepkgs=...", "exclude=..." ad now go do > it yourself (and sorry no example), obviously does not > help that much to users, and its CentOS's loss as well, > users go away to other distros, and ultimately many of > them are lost in the jungle. -- Bright Star > (Bry8Star). But you appear to be missing the "C" part in CentOS (or Community Enterprise OS). If you can contribute to the Wiki, then the immediate problem is solved in that such threads can be pointed to the Wiki and slows the growth of my CentOS list folder!

Frankly, if you have a good point to make that would benefit the masses and you have spare time, then it's best to create a Wiki page for it.

Cheers, ak.

There is already a fine page on this subject on the wiki.

http://wiki.centos.org/AdditionalResources/Repositories

The point being made is that various people have the knowledge and experience to advise a startup setting for priority= for each repo I know that what I'm using has caused conflicts that have been quite time consuming to resolve - what works for others would be most helpful I do recognise that this will vary depending upon what tools are required but as a start: 1. developer workstation - what repos and what priority 2. LAMP server - probably just CentOS repos and something which deals with later php / perl / ruby 3. web / internet workstation - needs audio and video stuff working just my thoughts for starters.

That totally depends on what you need to install and what repo it is in. Since 3rd party repos are constantly adding new packages that they did not have last week, it is impossible to say what would be the proper priorities.

I already posted what I personally do, which is:

Install CentOS and set Base, updates, extras, and fasttrack to a Priority=1

I usually do not need to enable centosplus, but if I do, I set it to Priority=2 and I put "excludes=<pkg_names>" in the Priority=1 repos for the packages I want let CentOS plus replace in those repos.

I then normally add EPEL and set the Priority=10 for that.

Hopefully, that is all I need to add.

If I have to add any more repositories, first make sure my packages are currently all updated by doing a yum upgrade. Then I add the new repos one at a time and I make them Priority=10 (the same as EPEL) ... and after I add them , i do a "yum update". If it tries to update, I look at the packages and decide if I am going to allow the update or not ... if I am ok to do the updates, then I do them and make sure it works. Then I would install the packages I need from that repo. Then I would add the next new repo till I get to the end.

The best scenario is that all your 3rd party repos can co-exist at the same Priority setting and that is where I start (at Priority=10) ... and if something does not work, I troubleshoot it and take individual action.

Each individual machine is going to require a unique and separate group of settings based on what you want to install ... which is why there is no official recommendations.

I personally am using the following repos right now on my main Desktop, which is CentOS-6.3:

adobe-linux-x86_64 | 951 B 00:00 base | 3.7 kB 00:00 cr | 3.0 kB 00:00 elrepo | 1.9 kB 00:00 elrepo-extras | 1.9 kB 00:00 extras | 3.5 kB 00:00 fasttrack | 3.5 kB 00:00 google-chrome | 951 B 00:00 google-musicmanager | 951 B 00:00 livna | 1.3 kB 00:00 nux-dextop | 2.7 kB 00:00 rpmforge | 1.9 kB 00:00 updates | 3.5 kB 00:00

All of the secondary repos are set to the same priorities and everything seems to work.

Right now I have an "exclude=wxGtk*" for rpmforge for some reason. And an "exclude=nx freenx*" for Nux! repo.

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

Les Mikesell wrote:

Using any single 3rd party repo is likely to work at least until the next update because the contents will have been tested against the base distro - as long as the packages are intended for use with RHEL/Centos.

I have a slightly different question. I had kbsingh-CentOS-Extras.repo enabled until recently. Is disabling a repo that has been in use likely to cause any problems?