Hi all,
I just subsribed to this list, but I know from the archive that this has
already been somewhat discussed on the list, still for completness first
a short intro.
For people outside the loop:
We a small group of FE contributers have been discussing creating /
instantitiating a FE security sig / team.
What we have sofar can be found on:
http://fedoraproject.org/wiki/Extras/Schedule/SecurityPolicy
The last 2 weeks it has been rather quiet in our little group I would
like to get the discussion on FE-security kickstarted again, hence this
mail.
To the people in the CC, afaik you're not subscribed yet, but you were
involved in the FE security discussion sofar. We initially commited to
taking this discussion public monday a week ago, well clearly we didn't.
So I'm taking it public through this list now and I would like todo the
rest if this discussion on this list, please subscribe.
To the people on the list please use reply to all so that those in the
CC stay involved in this thread.
After this intro hopefully everybody knows what I'm talking about / is
up2date, so now lets look forward.
My proposal to get an Fedora Extra Security Team on the road is as follows:
Fesco will discuss:
http://fedoraproject.org/wiki/Extras/Schedule/SecurityPolicy
Coming Thursday, hopefully with some improvements but if nescesarry as
is. I know that gives us just a few days to discuss any improvements,
but things have already been widely discussed and after that we've all
been quiet for a while. So I think its about time to take this to the
next level.
All in favor of getting this on the FESco speaking schedule soon say I
:) I ofcourse vote for my own proposal.
So we need to get:
http://fedoraproject.org/wiki/Extras/Schedule/SecurityPolicy
in tip-top shape before thursday. So what suggestions have come up sofar:
---
Josh bressers wrote:
"I've looked that document over in the past. I admit the times at the end
chart scare me. That's a fairly complicated chart. Within Red Hat there
was discussion about how to best classify security issues, this is what we
came up with:
http://www.redhat.com/security/updates/classification/
When one has to classify security threats, less is more.
I would suggest something more along these lines:
Critical: Don't bother waiting for the maintainer, do whatever it takes to
fix it.
Important: A few days.
Moderate: A few weeks.
Low: A few months."
I agree that its a good idea to use the RedHat security team
classifications. Anyone feel like updating the wiki (I'm low on time)?
About the suggested response time I join sides with Jason that their
should be a response time for Critical bugs, not automatic take-over by
the FE security team.
Also I think the times should be shorter then suggested by Josh, we're
talking about ping times here, not time till fix. Maybe we need another
word here. The biggest problem sofar is people who have been dead quiet
in bugzilla. So if I say the security team takes over if their is no
response within a week, I mean no response _at all_ if the maintainer
says yip that looks like a problem I'll look into it, then he has
responded and the response timer gets reset. so in this case as long as
a maintainer makes an entry about his progress every week all is ok and
the FE security team does not step in. The team could ofcourse offer
help suggest fixes, but we won't step in and push a fix, that is left to
the maintainer.
---
In general one of things which needs updating in our proposal the most
is that it should be made very clear that the FE security team is a
fallback and a fallback only. Normally the maintainers are 100%
responsible for the security updates for their own packages (for as far
as a volunteer can be responsible, the should feel 100% responsible.)
Can a native English speaker put something like this in their in very
strong yet friendly words?
---
Besides the response time and the making very clear that security is the
maintainers responsibility not the security teams we still need to work
out the Open issues list. As I've suggested before:
-I would like to suggest to send announcement to the list (and in the
same format) where FC security announcements get send, Josh is this
possible, can we get direct access, or maybe through you/ the whole
RH-security team?
-The FE security team needs a way to get involved in bugs / fixes where
all the info is under embargo. Again Josh, can you/ the whole
RH-security team play a role here? We ofcourse only need to be in the
loop if a package within FE has a hole.
-I've used the word FE security team instead of SIG above because I
think to the outside team sounds a lot better (professional) then SIG,
and this well help in being taking serious by the outside world (for
embargos for example) this has 2 disadvantages however:
*maintainers could get the idea that the team is responsible for the
security fixes, which its not they (the maintainers) are
*confusion with the redhat security team
So I'm not sure which name is better team or sig.
Thanks for your time reading this and please give your much valued opinion.
Regards,
Hans
