Re: Two new releases

On Thu, 2002-06-27 at 14:07, Michael Redinger wrote:
 
> It would be nice to have something as powerful as the Red Hat Satellite
> Sever, but free, of course ... *ducks and runs*
> 
> To be more specific, what I'm really looking forward to (we're currently
> evaluating the 1.3 series) is:
> - - adding RPMs dynamically without the need to rebuild the current 
>   database (or at least make it way faster).
> - - rhnsd support
>   This is really a big issue that will definitely take quite some time
>   if done properly.
> - - some kind of client tracking (honestly, I didn't really have a look
>   at this, the real work is done by one of my fellow employees).
>   "Simple" things like the patch level of clients etc. ...

This is all on the roadmap in the sql series. (later on though)

1.1 was supposed to be the first cut of sql, basically just replacing
the existing shelves with a real sql database (we picked postgres cuz
thats what *I* had the most experience with).

But we increasingly encountered greater and greater need at our (john
and i) respective sites for an anonymous  service (basically 1.0) that
would scale WAY WAY up.

We sidelined 1.1 and started 1.3 as the real development version. Thats
why there'll never be a 1.2 release.

The plan for now is 

1.3 - get apache/mod_python working really well
      fast as it can reasonably be, without getting into bizarre configs
      release this as the 1.4 series when its stable.
      Note that 1.3/1.4 will still use the python shelves.

1.5 - pick the 1.1 tree back up. Not to bad a forward port - while I 
     certainly didn't do it perfectly, current is divided into a 
      frontend/backend kind of split.

      The planned functionality here is SIMPLY to replace the shelves 
      with sql  - so there wouldn't be any new features, but just a 
      a tested way to migrate to a "real" database (john and I are both
      new to db programming - the patch was created by Toby Reaves)

      1.6 is the stable release of 1.5

1.7 - extend the 1.5 sql usage to include what I normally just call the 
      "client data". Start tracking machines, admins, who's got what
      package, etc etc etc.

      that becomes 1.8

It gets fuzzy, even in my head, after that.

2.0/2.1 might be were we start trying to have a web based or other 
configuration dohicky like the RHN web pages. Thats a lot of work, needs
full 1.7/client support, and isn't something I'm going to worry about
(all my web pages are fantastically ugly).
    

> 
> Talking about what I'd like to see in current - well, the comparison with
> the Red Hat Satellite Server isn't too much off target. Red Hat is doing
> a good job here, they just didn't realize (for now) that something like
> that may be useful to companies/universities/... with less than 1000 
> hosts (500+ doesn't seem to be enough, the salesperson didn't even seem
> to understand why I'd want to use their product ...).  They didn't even
> realize that 10.000 USD is quite some money, you'd have to spend some
> more money which probably renders RHSS unusable for most of us ...

at the 1.7/1.8 point in the roadmap, current would be a fair replacement
for satellite.

(Any RH people that read that will scoff, and rightly, but its not too
far off the mark)

> 
> Enough rant - what I'm trying to express: current is definitely taking the
> right approach (using RHAT's protocols and their clients and focusing 
> on the server), that's why we'll switch from the apt RPM port for the clients
> (and possibly autoupdate for the servers) to current - we simply think
> that current (while not being much better than apt for now) could
> get something more powerful.
> Thanks for this great product - we just hope it'll get even better ...

While I certainly am NOT going to support/ work on two development
releases at once (one stable and one devel is certainly plenty) download
and take a look at the 1.1 release. Its still on the ftp site (and from
our bitkeeper repository at bk://stenia.biology.duke.edu/current-1.1 )

> 
-- 
Hunter Matthews                          Unix / Network Administrator
Office: BioScience 145/244               Duke Univ. Biology Department
Key: F0F88438 / FFB5 34C0 B350 99A4 BB02  9779 A5DB 8B09 F0F8 8438
Never take candy from strangers. Especially on the internet.