Thank you for pointing this out, dims. I didn't notice this process of openstack. But i wonder how do you find the relationship between that bot's commit and the global requirements commit.
And sileht, from this commit https://github.com/openstack/requirements/commit/c7f69afd6af56e8f7956c6fa0bea8fd776151fe6, there seems not a robust reason to require kombu >= 3.0.7. It's because ubuntu Trusty's default kombu version is 3.0.7, and it works with amqp 1.4.0, right? On Wed, Oct 28, 2015 at 12:46 PM, Davanum Srinivas <dava...@gmail.com> wrote: > "Bot with No reason" <<< Not really accurate. The process in openstack is > to update global requirements first and then bot proposes the update to > different projects. So please look at > > https://github.com/openstack/requirements/commit/c7f69afd6af56e8f7956c6fa0bea8fd776151fe6 > for the commit which updated the global requirements > > Thanks, > dims > > On Wed, Oct 28, 2015 at 1:07 PM, me,apporc <appleorchard2...@gmail.com> > wrote: > >> Thanks again. >> >> This kombu >=3.0.7 requirement is added in commit >> 5b9fb6980220dbfa18bac4c3231d57efb493ebf0, which is from a Bot with no >> reason. >> >> As i see, we are directly requiring amqp >=1.4.0 in requirement.txt from >> commit 0c954cffa2f3710acafa79f01b958a8955823640 on. >> So maybe there is no need to require kombu >= 3.0.7 ? >> >> Another reason I am asking about this, in the latest centos 7 epel repo, >> the versions of these two packages are : >> python-kombu : 2.5.16-1.el7 >> python-amqp : 1.4.5-1.el7 >> >> I have not test whether there are problems about heartbeat with this >> version pair though. >> >> On Tue, Oct 27, 2015 at 11:02 PM, Mehdi Abaakouk <sil...@sileht.net> >> wrote: >> >>> >>> [1] seems just socket timeout issue, and admins can adjust those kernel >>>> params themselves. >>>> >>> >>> Yes, but if you trick kernel settings, like putting very low tcp >>> keepalive values, you don't need to have/enable heartbeat. >>> >>> [2] and [3] truly a problem about the heartbeat implementation, but it >>>> says >>>> the fix is a part of py-amqp 1.4.0, and the dependency with kombu was >>>> not >>>> specified. >>>> [4] is a bug of kombu's autoretry method which is said to be fixed in >>>> kombu >>>> 3.0.0, it is not directly related to heartbeat. >>>> >>> >>> As far as I can remember, this is because oslo.messaging doesn't really >>> require py-amqp but only kombu, so to ensure kombu depends of py-amqp 1.4.0 >>> we have to depends on kombu 3.0.7 (that have amqp>=1.4.0 in its >>> requirements I guess). >>> >>> >>> --- >>> Mehdi Abaakouk >>> mail: sil...@sileht.net >>> irc: sileht >>> >> >> >> >> -- >> Regards, >> apporc >> >> __________________________________________________________________________ >> OpenStack Development Mailing List (not for usage questions) >> Unsubscribe: >> openstack-dev-requ...@lists.openstack.org?subject:unsubscribe >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev >> >> > > > -- > Davanum Srinivas :: https://twitter.com/dims > > __________________________________________________________________________ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > > -- Regards, apporc
__________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev