Add cluster-wide indication for network instance of cluster - #150
Conversation
- It can distinguish the device related network instance or the cluster wide network instance, maybe needed for the native kubernetes applications usage Signed-off-by: naiming-zededa <naiming@zededa.com>
…of cluster - proto generated files. Signed-off-by: naiming-zededa <naiming@zededa.com>
| // logical network instance, which is required e.g. for app instances that | ||
| // may be relocated between cluster nodes. This may be useful for native | ||
| // kubernetes applications using the network instance. | ||
| bool cluster_wide = 53; |
There was a problem hiding this comment.
Does cluster_wide mean that each device gets its own isolated replica of the network instance (with a different UUID but the same configuration)? If so, I'm not sure cluster_wide is the best name, since the network instance doesn't actually provide connectivity between applications running on different devices.
There was a problem hiding this comment.
No. this will be no different from today's ENC, same UUID to all the nodes. But this is preparing for native kubernetes applications, which does not have any application reference to detail of NI inside pillar. This is to make a difference from device only NI, like those default NI generated.
Should we submit this even if we do not know yet if it is needed? We already have enough unused/deprecated fields from the past as it is. |
I think even w/ the current cluster NI implementation, it is good to know which NI downloaded is 'cluster-wide', today we don't have this. |
3 participants
wide network instance, maybe needed for the native kubernetes
applications usage