I agree. One could argue that people want to do this and are going to try so we may as well share what we know about doing this, including the warnings.
I will work with Aleksandr to get the details. I will write this up in the Operations Guide and include a link from the Planning Guide section about hardware requirements. https://bugs.launchpad.net/fuel/+bug/1375965 On Tue, Sep 30, 2014 at 9:38 AM, Roman Alekseenkov < ralekseen...@mirantis.com> wrote: > I think we should let users shoot themselves in a leg, but there should be > a big warning thrown into their face before they do so. Just discussed that > with Chris yesterday > > Roman > > On Tue, Sep 30, 2014 at 5:47 AM, Andrey Danin <ada...@mirantis.com> wrote: > >> AFAIR, we created this restriction because VM instances may eat all CPUs >> and then there will be a lot of problems with DB performance. We just >> prohibited a user from doing such "shoot-in-leg" configuration. We suggest >> to use a number of virtual machines as a demo/playground environment >> instead. >> >> On Mon, Sep 29, 2014 at 6:43 AM, Aleksandr Vorobiov < >> avorob...@mirantis.com> wrote: >> >>> Hello everyone, >>> >>> Recently I've been playing and trying to install MOS in "all in one" >>> mode, i.e. all services should be setup in one node. To accomplish this >>> I've edited the file >>> /usr/lib/python2.6/site-packages/nailgun/fixtures/openstack.yaml in nailgun >>> docker container. I've removed instructions which describe conflicts >>> between roles: >>> >>> ---------------------------------------------------------------- >>> controller: name: "Controller" description: "The controller initiates >>> orchestration activities and provides an external API. Other components >>> like Glance (image storage), Keystone (identity management), Horizon >>> (OpenStack dashboard) and Nova-Scheduler are installed on the controller >>> as well." conflicts: - compute >>> --------------------------------------------------------------------------------------------------- >>> >>> Strings in red were removed. >>> >>> After that I was able to assign controller, cinder and compute roles on >>> a single node. After finishing deployment I was able to complete OSTF tests >>> and checked installation manually, at first glance everything works well. >>> >>> The question is: is this behaviour expected and what issues could be >>> faced in such setup? >>> >>> Thank you in advance. >>> >>> -- >>> Mailing list: https://launchpad.net/~fuel-dev >>> Post to : fuel-dev@lists.launchpad.net >>> Unsubscribe : https://launchpad.net/~fuel-dev >>> More help : https://help.launchpad.net/ListHelp >>> >>> >> >> >> -- >> Andrey Danin >> ada...@mirantis.com >> skype: gcon.monolake >> >> -- >> Mailing list: https://launchpad.net/~fuel-dev >> Post to : fuel-dev@lists.launchpad.net >> Unsubscribe : https://launchpad.net/~fuel-dev >> More help : https://help.launchpad.net/ListHelp >> >> > > -- > Mailing list: https://launchpad.net/~fuel-dev > Post to : fuel-dev@lists.launchpad.net > Unsubscribe : https://launchpad.net/~fuel-dev > More help : https://help.launchpad.net/ListHelp > >
-- Mailing list: https://launchpad.net/~fuel-dev Post to : fuel-dev@lists.launchpad.net Unsubscribe : https://launchpad.net/~fuel-dev More help : https://help.launchpad.net/ListHelp