Is it ok if i create undercloud on VMware because its easier to take snapshot and clone etc.
~S On Thu, Feb 1, 2018 at 11:38 AM, Fawaz Mohammed <fawaz.moh.ibrah...@gmail.com> wrote: > In case your undercloud is dead, there will no be any impact in your > overcloud, but you will not be able to perform day 2 operations on your > tripleO environment, such as scale-out / scale-in the infrastructure, update > / upgrade your overcloud. > For that, it's recommended to backup the undercloud at least after each > operation in it. > > --- > Regards, > Fawaz Mohammed > > > On Feb 1, 2018 8:23 PM, "Satish Patel" <satish....@gmail.com> wrote: > > You have a point there! Got it, > > One more question what will happened if deploy my cloud and second day > my undercloud (TripleO) node is dead? is it going to impact my > overcloud production? because all recipe cooked inside undercloud and > now its no more. > > Do you think i need to keep backup of undercloud? > > On Thu, Feb 1, 2018 at 11:21 AM, Fawaz Mohammed > <fawaz.moh.ibrah...@gmail.com> wrote: >> On TripleO deployment, it's a challenge to use VMs for any overclode role >> (controller, compute or storage). As undercloud uses BMC / IPMI to control >> the power if bare metal during the deployment. >> >> --- >> Regards, >> Fawaz Mohammed >> >> >> On Feb 1, 2018 8:07 PM, "Satish Patel" <satish....@gmail.com> wrote: >> >> Thanks for info i will keep in mind, Can we use VM for controller >> node? or 1 physical and 2 VM controller node, just a thought >> >> On Thu, Feb 1, 2018 at 10:45 AM, Fawaz Mohammed >> <fawaz.moh.ibrah...@gmail.com> wrote: >>> For TripleO deployment, you'll need: >>> A separate node or VM as UnderCloud. >>> 3 control nodes for HA, 2 controllers can't form HA by default, this is >>> to >>> avoid split-brain scenarios. >>> For compute nodes, I recommend to have 2 x 10G ports, one for O&M and the >>> other one for tenants. And a separate 1G for BMC / IPMI. >>> >>> --- >>> Regards, >>> Fawaz Mohammed >>> >>> >>> On Thu, Feb 1, 2018 at 7:32 PM, Satish Patel <satish....@gmail.com> >>> wrote: >>>> >>>> We are CentOS7 shop that is why i have started playing with TripleO on >>>> my VMware environment to get familiar. what do you suggest for 20 node >>>> cluster? we don't have any storage requirement all we need CPU and >>>> Memory to run our application, currently all application running on >>>> VMware and i want to migrate them over Openstack and do some rapid >>>> deployment. >>>> >>>> This is what i am planning (let me if i am wrong) >>>> >>>> 2 controller node (in HA) >>>> 2 Cisco ASR1k (L3 function in HA) >>>> 20 compute node with 10G nic and lots of memory and many cpu cores. >>>> >>>> Anything else which i am missing? >>>> >>>> On Thu, Feb 1, 2018 at 10:15 AM, Fawaz Mohammed >>>> <fawaz.moh.ibrah...@gmail.com> wrote: >>>> > Hi Satish, >>>> > >>>> > TripleO (Supported on CentOS and RHEL) and Juju (Supported on Ubuntu) >>>> > [1] >>>> > are the most used OpenStack deployment tools. >>>> > >>>> > From the deployment perspectives, you need to know your cloud setup >>>> > and >>>> > the >>>> > required plugins to decide which deployment you will choose. As each >>>> > tool >>>> > has a pre-integrated plugging out of the box. >>>> > >>>> > >>>> > [1] >>>> > >>>> > >>>> > >>>> > https://docs.openstack.org/charm-deployment-guide/latest/install-juju.html >>>> > --- >>>> > Regards, >>>> > Fawaz Mohammed >>>> > >>>> > On Thu, Feb 1, 2018 at 5:04 PM, Satish Patel <satish....@gmail.com> >>>> > wrote: >>>> >> >>>> >> Hi Fawaz, >>>> >> >>>> >> Great so you think I'm doing right thing to use ASR1k for my L3 >>>> >> function? >>>> >> >>>> >> I'm using DVR on my lab with 3 node cluster and it's kinda >>>> >> complicated >>>> >> to >>>> >> troubleshoot sometime. >>>> >> >>>> >> We are planning to build 20 node cluster for start and started >>>> >> playing >>>> >> with tripleo to deploy and having lots of issue with tripleo too :( >>>> >> I >>>> >> was >>>> >> also thinking to use mirantis but they stopped development. >>>> >> >>>> >> Based on your experience what most of folks use to deploy middle size >>>> >> cloud? What about you? >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> Sent from my iPhone >>>> >> >>>> >> On Jan 31, 2018, at 11:48 PM, Fawaz Mohammed >>>> >> <fawaz.moh.ibrah...@gmail.com> wrote: >>>> >> >>>> >> http://www.jimmdenton.com/networking-cisco-asr-install/ can beet your >>>> >> requirements. I have another plugin in production from different >>>> >> vendor >>>> >> for >>>> >> the same purpose. and it works perfect. >>>> >> >>>> >> Regarding your question about the license, usually, there is no >>>> >> license >>>> >> for such plugins. >>>> >> >>>> >> I've no production experience with DVR, and I don't recommend it in >>>> >> medium >>>> >> to large environment. >>>> >> >>>> >> On Thu, Feb 1, 2018 at 2:58 AM, Satish Patel <satish....@gmail.com> >>>> >> wrote: >>>> >>> >>>> >>> What about this? >>>> >>> http://www.jimmdenton.com/networking-cisco-asr-part-two/ >>>> >>> >>>> >>> ML2 does use ASR too, >>>> >>> >>>> >>> Just curious what people mostly use in production? are they use DVR >>>> >>> or >>>> >>> some kind of hardware for L3? >>>> >>> >>>> >>> >>>> >>> On Wed, Jan 31, 2018 at 4:02 PM, Fawaz Mohammed >>>> >>> <fawaz.moh.ibrah...@gmail.com> wrote: >>>> >>> > Hi Santish, >>>> >>> > >>>> >>> > In my knowlege, Cisco has ml2 driver for Nexus only. >>>> >>> > >>>> >>> > So, if you have requirements for dynamic L3 provisioning / >>>> >>> > configuration, >>>> >>> > it's better to go with SDN solution. >>>> >>> > >>>> >>> > On Jan 31, 2018 11:39 PM, "Satish Patel" <satish....@gmail.com> >>>> >>> > wrote: >>>> >>> >> >>>> >>> >> So no one using ASR 1001 for Openstack? >>>> >>> >> >>>> >>> >> On Tue, Jan 30, 2018 at 11:21 PM, Satish Patel >>>> >>> >> <satish....@gmail.com> >>>> >>> >> wrote: >>>> >>> >> > Folks, >>>> >>> >> > >>>> >>> >> > We are planning to deploy production style private cloud and >>>> >>> >> > gathering >>>> >>> >> > information about what we should use and why and i came across >>>> >>> >> > with >>>> >>> >> > couple of document related network node criticality and >>>> >>> >> > performance >>>> >>> >> > issue and many folks suggesting following >>>> >>> >> > >>>> >>> >> > 1. DVR (it seem complicated after reading, also need lots of >>>> >>> >> > public >>>> >>> >> > IP) >>>> >>> >> > 2. Use ASR1k centralized router to use for L3 function (any >>>> >>> >> > idea >>>> >>> >> > what >>>> >>> >> > model should be good? or do we need any licensing to integrate >>>> >>> >> > with >>>> >>> >> > openstack?) >>>> >>> >> > >>>> >>> >> > Would like to get some input from folks who already using >>>> >>> >> > openstack >>>> >>> >> > in >>>> >>> >> > production and would like to know what kind of deployment they >>>> >>> >> > pick >>>> >>> >> > for network/neutron performance? >>>> >>> >> >>>> >>> >> _______________________________________________ >>>> >>> >> Mailing list: >>>> >>> >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack >>>> >>> >> Post to : openstack@lists.openstack.org >>>> >>> >> Unsubscribe : >>>> >>> >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack >>>> >> >>>> >> >>>> > >>> >>> >> >> > > _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
