Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

How much sense has magnum implementation on RDO Ocata via packstack ?

Just a sample - docker swarm VMs are using cinder volumes to spawn. But, cinder services due to known restriction (CI related) may run only on Controller. Due to well known packstack limitation as of Newton Release ( not to split Storage Node anymore ) . Thus answer appears unfortunately -
"Just for fun" . Please, correct me If I am wrong about that. A bunch of people are still using packstack for RDO deployments. Might be it makes some sense to create "Infra" ( vs it was stated by J.P. on rdo mailing list ).

How much sense has magnum implementation on RDO Ocata via packstack ?

Just a sample - docker swarm VMs are using cinder volumes to spawn. But, cinder services due to known restriction (CI related) may run only on Controller. Due to well known packstack limitation as of Newton Release ( not to split Storage Node anymore ) . Thus answer unfortunately appears unfortunately -
to be "Just for fun" . Please, correct me If I am wrong about that. A bunch of people are still using packstack for RDO deployments. Might be it makes some sense to create "Infra" ( vs it was stated by J.P. on rdo mailing list ).

How much sense has magnum implementation on RDO Ocata via packstack ?

Just a sample - docker swarm VMs are using cinder volumes to spawn. But, cinder services due to known restriction (CI related) may run only on Controller. Due to well known packstack limitation as of Newton Release ( not to split Storage Node anymore ) . Thus answer unfortunately appears to be "Just for fun" . Please, correct me If I am wrong about that. A bunch of people are still using packstack for RDO deployments. Might be it makes some sense to create "Infra" ( vs it was stated by J.P. A.P. on rdo mailing list ).

How much sense has magnum implementation on RDO Ocata via packstack ?

UPDATE
I still believe that creating storage.pp role been able to pass CI and responsible for separation Swift,Cinder,Glance services from Controller set up via packstack is not the task out of scope for RH Cloud team. Changes been done in Newton release of RDO will cause performance degradation on production system making containers (kubernetes,docker swarm) support on RDO Ocata almost impossible.
TripleO has serious issues with maintenance of production environment. Wrong allocation memory per CPU causes heat update to crash even during attempt to add compute node to overcloud. Original deployment of TripleO may be redone as many times as needed versus failure of redeploying ( heat update of overcloud stack ) overcloud as entity during maintenance either recovery HA Controller been crashed .
The question - is it worse of efforts to maintain packstack responsible for containers support via magnum or no ?
END UPDATE

Just a sample - docker swarm VMs are using cinder volumes to spawn. But, cinder services due to known restriction (CI related) may run only on Controller. Due to well known packstack limitation as of Newton Release ( not to split Storage Node anymore ) . Thus answer unfortunately appears to be "Just for fun" . Please, correct me If I am wrong about that. A bunch of people are still using packstack for RDO deployments. Might be it makes some sense to create "Infra" ( vs it was stated by A.P. on rdo mailing list ).

How much sense has magnum implementation on RDO Ocata via packstack ?

UPDATE
I still believe that creating storage.pp role been able to pass CI and responsible for separation Swift,Cinder,Glance services from Controller set up via packstack is not the task out of scope for RH Cloud team. Changes been done in Newton release of RDO will cause performance degradation on production system making containers (kubernetes,docker swarm) support on RDO Ocata almost impossible.
TripleO has serious issues with maintenance of production environment. Wrong allocation memory per CPU causes heat update to crash even during attempt to add compute node to overcloud. Original deployment of TripleO may be redone as many times as needed versus failure of redeploying ( heat update of overcloud stack ) overcloud as entity during maintenance either recovery HA Controller been crashed .
The question - is it worse worth of efforts to maintain packstack responsible for containers support via magnum or no ?
END UPDATE

Just a sample - docker swarm VMs are using cinder volumes to spawn. But, cinder services due to known restriction (CI related) may run only on Controller. Due to well known packstack limitation as of Newton Release ( not to split Storage Node anymore ) . Thus answer unfortunately appears to be "Just for fun" . Please, correct me If I am wrong about that. A bunch of people are still using packstack for RDO deployments. Might be it makes some sense to create "Infra" ( vs it was stated by A.P. on rdo mailing list ).