[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: What's wrong with debian?



Greetings,...

Am Dienstag, 1. März 2005 00:51 schrieb Eric Gaumer:
> Robert Brockway wrote:
> > On Mon, 28 Feb 2005, Eric Gaumer wrote:
> >>I don't really understand peoples concern. If Sarge was released
> >>tomorrow it would be the same group of packages sitting there today
> >>(relatively speaking).
> >>
> >>My point is if you think Sarge is stable (which it is) then go ahead and
> >>upgrade/install. What difference do the politics play? Security fixes
> >
> > It's not the politics - its the policy.  Debian approaches Stable and
> > Testing releases differently.
> >
> > Can I apt-get update && apt-get upgrade a Sarge box with the same
> > confidence that nothing will break that I put into Woody?  If so then
> > this needs to be stated because AFAIk Sarge is not even in code freeze
> > yet - something might break badly at the next upgrade.  I can't take this
> > risk in production.  Also, security updates are not formally applied
> > seperately to the Testing distro - this is also a big deal for a box in

>
> Well we have over 300 Debian servers in production most running Sarge. So
> you claim that things could break badly during an upgrade (and that seems
> to be everyones argument) but I can tell you that if you are not careless
> then things wont break.

Nice to hear so, but this isn't the point.

> Check the bug reports before upgrading some some major piece of software
> that you fully rely on. We have our own Debian repository where we refine
> packages that need it and package software that isn't part of Debian (some
> proprietary). We have special test environments where we can evaluate the
> system software. It's just a matter of your staff's talent.

and of the available ressources. If your team is managing 300 servers, you are 
propably using your own debian distribution, at least you say so. I have 
neither time nor I get paid for doing so.
Furthermore debian is an end-user product and not just a bunch of packets 
everybode may use as he or likes to do.
From that point, it is also not a matter of the staff's talent

> Nothing comes cost free. Debian does a very good job of providing three
> stable pools of software to pull from (yes three). In the end, it's your
> talent that dictates which one you will use. If they (Debian) rushed things
> out the door or made haste decisions then people would complain about bugs.
> They take the time to "get it right" but people complain things move to
> slowly. Bottom is you can't please everyone.

Well, of course you cannot please everyone. A lot of debian users were 
satisfied with a release every two years or get used with it.
Observing sarge's release process and progress it comes into my mind, that 
sarge actually contain two releases in one. Sarge seems to introduce a lot of 
new stuff into debian, enough for two releases. Perhaps Sarge is just too fat 
in that way.

> It would be a different story if they only offered Woody. That simply isn't
> the case. You have the choice to decide what you want. You want the newest
> stuff yet you want it guaranteed free of problems. If anyone thinks this a
> small burden then I recommend you give it a try. If it was that easy you
> would all be running Sarge and "stabilizing" it on your own. It's not that
> easy because I deal with it daily.

Well, running Sarge is not that easy. If you don't have a huge testing 
enviroment, multiple fallback-Servers it's quite risky.

> For instance, we have an LDAP server that houses about 6000 users for a
> Postfix mail server. It's been having serious issues due the select()
> system call limitations. New versions (upstream) have a compile time option
> that enables many more connections. I spent an entire weekend re-packaging
> slapd and all of its dependencies so that they would integrate into our
> Debian environment.

Brave for you for doing so, but most of us don't.

> It does me no good to complain about why newer packages aren't available. I
> get paid to make things happen and I appreciate all the work the Debian
> maintainers do to make my life easier so the least I could do is handle
> certain issues on my own. Isn't that what open source is all about?

You get paid for doing so - this is a good by point, but most companies don't 
pay and don't want this.

> If this still isn't making sense then get involved with the Debian and see
> just how much stuff is involved with a new release. Most just take these
> things for granted.

I don't say there is no new stuff involved. Progress is measureable, that's 
what I said. But in my concerns it's just too much and justifies two 
releases. 

> Sorry... didn't mean to ramble. I just wish everyone would redirect their
> angst of releasing Sarge toward helping it's release rather complaining
> about its delay. You don't need to be a developer to report a bug. 

I know, and I'm reporting bugs.
Well, to be honest, I've been spending more time on testing FreeBSD in our 
environment than testing, tuning and using sarge for the last 1-2 months.

> If you 
> have a production environment then setup a test machine to track potential
> problems. This would help speed up new releases...

Of course, but somehow I resign. I'm doubt whether it is justified to spend 
time into debian and not migrating to an alternative unix/linux distro.

Keep smiling
yanosz



Reply to: