Currently all plugins run within the same node process as hoodie-server. In the future it should be possible to run plugins in separate node processes that run on the same or even different machines.
The plugin code should not have to change for that. hoodie-pllugins-manager should just distribute jobs over available workers. In addition, maybe we want to have auto-launching of new workers if loads increases.
For now this can all be done using the hoodie.task API and implementation, but once this reaches a high-traffic situation for a single app, we might have to look into proper queues to implement hoodie.task behind the scenes.
Currently all plugins run within the same node process as
hoodie-server
. In the future it should be possible to run plugins in separate node processes that run on the same or even different machines.The plugin code should not have to change for that.
hoodie-pllugins-manager
should just distribute jobs over available workers. In addition, maybe we want to have auto-launching of new workers if loads increases.For now this can all be done using the
hoodie.task
API and implementation, but once this reaches a high-traffic situation for a single app, we might have to look into proper queues to implementhoodie.task
behind the scenes.