On Wed, Oct 2, 2019 at 9:59 PM Brian Nixon <[email protected]> wrote:

> NIO is still the default server factory so I'm guessing many users of 3.4
> simply aren't configuring Netty. And our recommendation for users who want
> Netty could be to upgrade to a 3.5 release as that should be better in
> every way for them.
>
> Is there a principle determining the difference between leaving the code
> available in 3.4 with a warning attached and removing the code entirely so
> that they would have to independently modify and package in order to use
> the feature?
>
>
Primarily/historically what i mentioned - we don't introduce major
features/changes (esp non-b/w compat) in fix releases.

Patrick


>
> On Wed, Oct 2, 2019 at 8:48 AM Patrick Hunt <[email protected]> wrote:
>
> > On Wed, Oct 2, 2019 at 1:49 AM Andor Molnar <[email protected]> wrote:
> >
> > > Hi Pat,
> > >
> > > Would you please clarify what do you mean “dropping netty support from
> > > 3.4”?
> > >
> > >
> > My simplistic thought was just that. Ship new versions of 3.4 that remove
> > support for netty. That could mean turning if off by default (not sure
> how
> > much work that would be) or just purging the netty code from the codebase
> > entirely. (3.4). It would be an exception to our "don't break b/w compact
> > in fix releases" policy, but this is an extreme case imo. We have no
> > intention of supporting netty in 3.4 going forward as evidenced by the
> fact
> > that the netty version is locked to netty 3 (long out of support by netty
> > as they are no longer backporting fixes) and we have no intention of
> > updating to the new version of netty on 3.4. Maybe this CVE don't affect
> > us, but at some point it will. Users have the option to move to a stable,
> > b/w compat, 3.5. release. Not optimal I agree.
> >
> >
> > > Does that mean we won’t submit security patches from now on, but keep
> the
> > > Netty classes (NettyServerCnxnFactory and ClientCnxnSocketNetty)
> > available
> > > OR remove these classes from the codebase?
> > >
> > > The latter means we’ll drop client SSL feature too.
> > >
> > >
> > Say there is a new CVE on netty and it's not backported to netty3, what
> > would we do in that case. I guess we could wait/kick the can down the
> road
> > till we really hit that. For the moment just say that it doesn't affect
> us
> > as you researched and add to 3.4 exceptions.
> >
> > This is just my suggestion/option rather than a recommendation, open to
> > other ideas. ;-)
> >
> > Patrick
> >
> >
> > > Andor
> > >
> > >
> > >
> > > > On 2019. Oct 2., at 2:27, Michael Han <[email protected]> wrote:
> > > >
> > > >>> How about officially dropping netty support from 3.4 and asking
> > people
> > > > to move to the new version
> > > > +1. This sounds a good opportunity to deprecate 3.4 branch.
> > > >
> > > > On Tue, Oct 1, 2019 at 8:00 AM Enrico Olivelli <[email protected]>
> > > wrote:
> > > >
> > > >> Il giorno mar 1 ott 2019 alle ore 16:15 Patrick Hunt <
> > [email protected]>
> > > ha
> > > >> scritto:
> > > >>
> > > >>> Another option/solution: How about officially dropping netty
> support
> > > from
> > > >>> 3.4 and asking people to move to the new version (3.5 stable or
> > later)?
> > > >>>
> > > >>
> > > >> Sounds good
> > > >>
> > > >> Enrico
> > > >>
> > > >>
> > > >>>
> > > >>> Patrick
> > > >>>
> > > >>> On Tue, Oct 1, 2019 at 4:22 AM Andor Molnar <[email protected]>
> > wrote:
> > > >>>
> > > >>>> I agree with 3.4 should not be refactored in any way even for a
> > > >> security
> > > >>>> fix.
> > > >>>>
> > > >>>> What's wrong with the "alpha story"?
> > > >>>>
> > > >>>> I think releasing in an early stage with "-alpha", "-beta"
> modifiers
> > > is
> > > >>>> not a bad thing alone, as long as it doesn't take years to get to
> > the
> > > >>>> stable release.
> > > >>>>
> > > >>>> Andor
> > > >>>>
> > > >>>>
> > > >>>> On Tue, 1 Oct 2019, Enrico Olivelli wrote:
> > > >>>>
> > > >>>>> Date: Tue, 1 Oct 2019 10:54:24 +0200
> > > >>>>> From: Enrico Olivelli <[email protected]>
> > > >>>>> Reply-To: [email protected]
> > > >>>>> To: [email protected]
> > > >>>>> Subject: Re: [VOTE] Apache ZooKeeper release 3.5.6 candidate 2
> > > >>>>>
> > > >>>>> Il mar 1 ott 2019, 10:38 Andor Molnar <[email protected]> ha
> > scritto:
> > > >>>>>
> > > >>>>>> Backporting Netty 4 would be a huge, cumbersome task, I hope we
> > > >> don’t
> > > >>>> have
> > > >>>>>> to do it.
> > > >>>>>>
> > > >>>>>
> > > >>>>> Yes, 3.4 is mature and stable and closed for refactors.
> > > >>>>>
> > > >>>>>
> > > >>>>>> However I had a quick look at the details of this CVE and it
> seems
> > > >> to
> > > >>> me
> > > >>>>>> that it only affects the HTTP codec:
> > > >>>>>>
> > > >>>>>>
> > > >>>>
> > > >>>
> > > >>
> > >
> >
> https://github.com/netty/netty/commit/39cafcb05c99f2aa9fce7e6597664c9ed6a63a95
> > > >>>>>>
> > > >>>>>> Can’t we just say 3.4.14 is not affected?
> > > >>>>>> We’re not running HTTP server inside ZooKeeper.
> > > >>>>>>
> > > >>>>>> Otherwise we might be able to release 3.6.0-alpha1 now, put a
> date
> > > >> for
> > > >>>> 3.4
> > > >>>>>> EOL and highlight on the webpage that this
> > > >>>>>>
> > > >>>>>
> > > >>>>> Please do not start an 'alpha' story like for 3.5....
> > > >>>>>
> > > >>>>> CVE probably won’t be resolved on that branch, please upgrade to
> > 3.5.
> > > >>>>>>
> > > >>>>>
> > > >>>>> +1
> > > >>>>>
> > > >>>>>
> > > >>>>> Enrico
> > > >>>>>
> > > >>>>>>
> > > >>>>>> As a third option we could ask Norman to kindly fix 3.10.6.Final
> > as
> > > >>>> well…
> > > >>>>>> or submit a PR ourselves, it doesn’t seem to me a big deal.
> > > >>>>>>
> > > >>>>>
> > > >>>>> Not so useful
> > > >>>>>
> > > >>>>>>
> > > >>>>>> What do you think?
> > > >>>>>>
> > > >>>>>> Andor
> > > >>>>>>
> > > >>>>>>
> > > >>>>>>
> > > >>>>>>
> > > >>>>>>> On 2019. Oct 1., at 2:00, Patrick Hunt <[email protected]>
> wrote:
> > > >>>>>>>
> > > >>>>>>> I pushed patches for 3.5 and trunk and the tests passed on my
> > mac.
> > > >>>>>> However
> > > >>>>>>> 3.4 is using netty 3.10.6.Final and as such it's not a simple
> > > >>> upgrade.
> > > >>>>>>> (there are no fixes against 3.10 for this CVE, at least not so
> > far)
> > > >>> Not
> > > >>>>>>> sure what we want to do about this... someone would need to
> > > >> backport
> > > >>>> the
> > > >>>>>>> netty 4.1 changes into 3.4 afaict.
> > > >>>>>>>
> > > >>>>>>> Patrick
> > > >>>>>>>
> > > >>>>>>> On Mon, Sep 30, 2019 at 1:08 PM Patrick Hunt <[email protected]
> >
> > > >>> wrote:
> > > >>>>>>>
> > > >>>>>>>> I'll work on it today.
> > > >>>>>>>>
> > > >>>>>>>> Patrick
> > > >>>>>>>>
> > > >>>>>>>> On Mon, Sep 30, 2019 at 11:59 AM Enrico Olivelli <
> > > >>> [email protected]
> > > >>>>>
> > > >>>>>>>> wrote:
> > > >>>>>>>>
> > > >>>>>>>>> Okay
> > > >>>>>>>>>
> > > >>>>>>>>> I am cancelling the release.
> > > >>>>>>>>>
> > > >>>>>>>>> I have a problem with my box, I can't work on netty upgrade.
> > > >>>>>>>>>
> > > >>>>>>>>> Any volounteer?
> > > >>>>>>>>>
> > > >>>>>>>>> Enrico
> > > >>>>>>>>>
> > > >>>>>>>>> Il lun 30 set 2019, 20:32 Andor Molnar <[email protected]> ha
> > > >>>> scritto:
> > > >>>>>>>>>
> > > >>>>>>>>>> The good news is: we need to release 3.4.15 too. :)
> > > >>>>>>>>>>
> > > >>>>>>>>>> Andor
> > > >>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>>> On 2019. Sep 30., at 20:26, Patrick Hunt <[email protected]
> >
> > > >>> wrote:
> > > >>>>>>>>>>>
> > > >>>>>>>>>>> created:
> > https://issues.apache.org/jira/browse/ZOOKEEPER-3563
> > > >>>>>>>>>>>
> > > >>>>>>>>>>> On Mon, Sep 30, 2019 at 11:20 AM Patrick Hunt <
> > > >> [email protected]>
> > > >>>>>>>>> wrote:
> > > >>>>>>>>>>>
> > > >>>>>>>>>>>> -1 - when I run dependency check on the release candidate
> > > >>> artifact
> > > >>>>>>>>> it's
> > > >>>>>>>>>>>> failing with:
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>> [ERROR] netty-transport-4.1.29.Final.jar: CVE-2019-16869
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>> I ran this on trunk and it's passing, as such it must be
> an
> > > >>> issue
> > > >>>>>>>>> with
> > > >>>>>>>>>> the
> > > >>>>>>>>>>>> the 3.5.6 netty version specifically. It's listed as a
> high,
> > > >> we
> > > >>>>>>>>> should
> > > >>>>>>>>>>>> patch this as well before releasing.
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>> Patrick
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>> On Sun, Sep 29, 2019 at 7:29 AM Enrico Olivelli <
> > > >>>>>> [email protected]
> > > >>>>>>>>>>
> > > >>>>>>>>>>>> wrote:
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>>>> This is a bugfix release candidate for 3.5.6.
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> It fixes 28 issues, including upgrade of third party
> > > >> libraries,
> > > >>>>>>>>>>>>> TTL Node APIs for C API, support for PCKS12 Keystores,
> and
> > > >>> better
> > > >>>>>>>>>>>>> procedure
> > > >>>>>>>>>>>>> for the upgrade of servers from 3.4 to 3.5.
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> The full release notes is available at:
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>
> > > >>>>>>
> > > >>>>
> > > >>>
> > > >>
> > >
> >
> https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12310801&version=12345243
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> *** Please download, test and vote by October 2nd 2019,
> > 23:59
> > > >>>>>> UTC+0.
> > > >>>>>>>>>> ***
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> Source files:
> > > >>>>>>>>>>>>>
> > > >>> https://people.apache.org/~eolivelli/zookeeper-3.5.6-candidate-2
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> Maven staging repo:
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>
> > > >>>>>>
> > > >>>>
> > > >>>
> > > >>
> > >
> >
> https://repository.apache.org/content/repositories/orgapachezookeeper-1042/
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> The release candidate tag in git to be voted upon:
> > > >>>>>> release-3.5.6-rc2
> > > >>>>>>>>>>>>>
> https://github.com/apache/zookeeper/tree/release-3.5.6-rc2
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> ZooKeeper's KEYS file containing PGP keys we use to sign
> > the
> > > >>>>>>>>> release:
> > > >>>>>>>>>>>>> https://www.apache.org/dist/zookeeper/KEYS
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>> Should we release this candidate?
> > > >>>>>>>>>>>>> Enrico Olivelli
> > > >>>>>>>>>>>>>
> > > >>>>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>>
> > > >>>>>>>>>
> > > >>>>>>>>
> > > >>>>>>
> > > >>>>>>
> > > >>>>>
> > > >>>
> > > >>
> > >
> > >
> >
>

Reply via email to