Revisor on Enterprise Linux

Although these are just preliminary results; Revisor (in CLI mode) now also runs on Enterprise Linux 5!

I've composed a minimal Fedora 7 installation CD which (while I'm writing this), is done formatting the partitions of a QEMU guest, and installing packages.

-some time later-
It's booting as well. In the meantime, I've been trying to get live media going on EL5 too, but I haven't managed to get it working, yet.

Tagged nonetheless and built: http://koji.fedoraproject.org/koji/packageinfo?packageID=4446

Spinning Fedora X, X-1 and X+1

In the past, I've been told building Fedora X would require the host doing the building to also run Fedora X. Not Fedora X+1 or Fedora X-1. Really Fedora X. It wouldn't be possible any other way. I assure you it wasn't my grandma or girlfriend telling me this.

Taking matters into my own hands, being the stubborn m-f that I am, I've been successfully building

on F7: FC6, F7 and F8t1 or rawhide
on F8t1 (rawhide): FC6, F7 and F8t1 or rawhide.

I've just not been able to build anything on X-1, but that's it. The rest, I did. And it's my very own application failing to do anything on FC6 because I've managed to create pretty hard requirements towards Fedora 7 or later.

Some of you think: "You didn't. It's impossible. I told you it was impossible. Period."

I'll add to that, that I've not only successfully built these spins -meaning that I didn't get any errors or warnings during the builds-, but they also successfully passed Q&A (although as we know FC6 had bugs in doing certain types of installations, so it didn't really /pass/ Q&A but it did anyway, and of course I'm not sure what Q&A requirements have been set for F8(t1)).

Bingo.

NetworkManager and NetworkManagerDispatcher

Imagine you connect your laptop to your home network and you want to authenticate against nis.home.org, using automatically mounted home directories from fileserver.home.org. The following morning, you arrive at the office and connecting your laptop to the network there, you wait for ypbind to bind to nis.home.org (which just so happens to be unavailable), and you cannot log in directly as your home directory cannot be mounted (because fileserver.home.org just so happens to be unavailable too). In addition, your office uses LDAP authentication and has a group data share you want to mount.

I'm guessing you'll need to at least disable ypbind, and maybe change authentication configuration and start using LDAP. You're also probably gonna want to mount that group data share and not automount your home directory from some NFS share.

Are you performing all these changes by hand? Does your office allow you to change it all? How does the office manage to support all that? I don't know.

What I do know is that I personally have a similar use-case with my laptop that I want to be able to connect to my home network having NIS and autofs, while not having to change the system configuration when I'm 'offline' -not connected to my home network.

I've begun to use the NetworkManagerDispatcher: it allows me to execute some scripts that would be able to detect wherever I am and adjust my system configuration accordingly. I'll call it NetworkManagerDispatcherScripts (don't try and say that out loud very fast, twice). There's a proof of concept piece of code at http://git.kanarip.com/?p=nmdscripts/.git which you can get with;

git clone git://git.kanarip.com/nmdscripts/

Execute the ./nmdscript script and see if you're connected to one of the pre-configured networks (which all apply to my home networks). Until now, it's just detecting some network characteristics and not executing any system configuration changes, and I'm just curious what anyone thinks about it.

Unluckiest person for RH302, ever

Originally, I had planned to take RH300 (RH302 including the Rapid Course Track) July 9th. When it seemed that a FUDCon was to be held starting July 13th (the day I would have taken the actual exam RH302), I canceled my seat and planned for a long stay at Raleigh, NC -which at that time was the location where FUDCon was to be held. Then, as you all know, FUDCon was moved to either 3-5 August, or 10-12 August, and I really hoped and lobbied for the first weekend, because I knew July 27th, there was a RH302 too. Instead of staying a week longer and then take the RHCE Exam, I'd have arrived a week earlier and took a jet-lag exam trying to pass for RHCE.

Back to the original point of the post; since all that was canceled, I tried to get back a seat in the RH300 course I had planned originally -July 9th. I succeeded ;-)

After a 4 day RHEL5 internal workings re-cap, I thought I was ready to take the exam. I'll tell you right-away, I didn't pass.

Half-way through the exam, the network failed. Not just my network, but the entire network. Obviously, that sucks. The exam got canceled but I gotta say: That shit just happens, there's nothing anyone could have done about it.

Anyway I had to re-take the exam at some later day and of course the sooner the resit, the better. Red Hat apparently had some trouble re-scheduling for today, July 27th, which would have been the first opportunity, if not the RH300 course track planned for that week had been canceled. They told us the next (sure) opportunity would be August 17th (that is a 5 week wait just to re-take the exam), and that they'd try to arrange something for July 27th.

Just now however I got an email from Farnborough asking me why I couldn't attend the resit today, because they seemed to have missed me. I didn't know there was a resit today. I didn't get any seat confirmation, invitation or notification, whatsoever. I didn't get any for August 17th either. I sure hope I get a chance to pass RHCE some time, hopefully without too much trouble ;-)

I seem to be the unluckiest person wanting to take the RH302 exam, ever ;-)

Developing in Fedora

You'll usually start developing some feature or application because you think you can create something or improve something else. You'll probably get excited and talk about it and at some point, come across the people that you thought know it all and will catch your drift and help you accomplish whatever it is you want. I've been proven wrong there. Mind that I'm developing a customization tool build upon existing release engineering tools, this doesn't hold up for entirely new tools.

The real test isn't the code. The real test isn't all the bugs you'll get either. Neither does it concern you, what huge pile of feature requests you'll get, or what number of bugs will show up once you try and implement some of these requested features, breaking other stuff in your code. Hell, there's no challenge in doing it 'enterprisey' either. The challenge you had in mind was to get things done, to move forward. That is what it is all about, and that just so happens to be what you're really good at.

Neither of all that is a real challenge, believe me.

The real challenge is to continue working as you see fit, because you'll get flamed from all sides that has people with a different perspective on things. The real challenge is to withstand the temptation of just quitting.

Continuing on building a customization application upon release engineering tools; I'm sure you can appreciate the different perspective on things. Whereas release engineering tools need to be stable and consistent, customization tools can just have their way with it. Don't overestimate the number of features you can add to a release engineering tool; it doesn't need all that, the tool works as is, so your life then is about discussing the need for this or that feature in some upstream tool, not about doing the code and solving the bugs. You'll give up eventually. You'll let them pull whatever they like from your tool.

This is the Fedora world, where everything moves forward fast, faster, fastest. It's a community distribution that does cutting-edge in stable, and for which enterprise linux engineers decide what progresses slow enough to keep track off changes, or what is too fast to really be allowed to be build upon whatever they are on the receiving end of bugs for.

Creating Re-Spins with Revisor

Creating re-spins has never been this easy. Of course we had pungi and livecd-tools to which we had added this GUI thing in Revisor, but since I've re-enabled the CLI mode in Revisor, re-spins go even easier and quicker by just copying these lines:

i386 Fedora 7 DVD:
revisor --cli --model=f7-i386 --dvd \
--kickstart=/etc/revisor/conf.d/fedora-7-gold.cfg

i386 Fedora 7 CD Set:
revisor --cli --model=f7-i386 --cd \
--kickstart=/etc/revisor/conf.d/fedora-7-gold.cfg

x86_64 Fedora 7 DVD:
revisor --cli --model=f7-x86_64 --dvd \
--kickstart=/etc/revisor/conf.d/fedora-7-gold.cfg

x86_64 Fedora 7 CD Set:
revisor --cli --model=f7-x86_64 --cd \
--kickstart=/etc/revisor/conf.d/fedora-7-gold.cfg

And if you want both CDs and DVDs, just append another command line switch! Isn't that just *amazing*?

I'm gonna add to allow more-then-one model from the CLI... Maybe it's useless, but it'll look awesome.

Revisor Enabling Translations

Revisor, having released it's final 2.0.3 version (2.0.3.10 is now in updates-testing), hopefully fixing any bugs in the feature set we had up and until now, has now moved to add all kinds of features to finally, at some point, become available to you in the 2.0.4.x version set.

We have not yet listed our priorities as to what feature we add first, but two things that were not in Revisor yet have always been quite important to Fedora; localization and accessibility.

I'm glad to announce that we've enabled Revisor to be fully localized, with help of Dimitris Glezos (Fedora l10n Team, helped me getting my ducks in a row) and Piotr "Raven" Drag (being the first one to translate Revisor to pl_PL). Now, obviously, we need translators to start translating ;-)

The other feature, accessibility, is harder, I think. If anyone of you that reads this can help us with that, I'd appreciate.

To get the latest and "greatest": http://revisor.fedoraunity.org/documentation/building-revisor-from-source/