[WBEL-users] Wondering about RedHat
Rafael Baquero S.
Thu, 30 Sep 2004 13:46:09 -0500
On Wednesday 29 September 2004 22:00, Paul wrote:
> On Wed, 2004-09-29 at 18:31, Rafael Baquero S. wrote:
> > I absolutelly agree with you.
> > Now I have one question I?
> > Should WB updates follow RHEL updates as soon as posible or should a time
> > period be allowed for RHEL updates to be tested? What would be correct
> > amount of time to wait for security updates, bug fixes, etc? Or maybe a
> > policy of double checking RHEL updates could be established?
> > Maybe the time has come for a Whitebox REL (Reliable Enterprise Linux).
> The issue with that is then it becomes the responsibility of somebody to
> dig though bugzilla looking for problems for a "reasonable time" after a
> patch is released to see if it causes issues, and that is not simple to
> The other way is to do two tier update-testing with patches being
> migrated to update-working after a couple weeks if no negative
> feedback. This probably would not catch that many issues because of the
> number & variety of systems running the testing updates would not be
> enough to find any but the most obvious bugs.
Posibly trying to partially automate the process of digging through bugzilla
looking for problems combined with a two-tier aproach for updating could do
The transfer of packages from update-testing to update-automatic could be done
by means of programs that check bugzilla for tickets for that particular
package-version. If any tickets are found the program checks the reports for
keywords (problem, crash, security hole, etc.) and either transfer the
packages to security updates if none of the keywords are found, or notify a
human to check the tickets before aproving the packages.
> Then there is the issue with what do you do with serious security
> related patches, to you put them though the same testing or do they get
> fed straight through?
The "quarantine" time for each update would have to be adjusted according to
the severity of the bug. This posibly could also be partialy/totally
> Paul Berger
> Whitebox-users mailing list