4 ms·

Could someone give me an example of when I would use it? It seems to me rather complicated by reading the text even though it promises a simple decentralized so

Could someone give me an example of when I would use it? It seems to me rather complicated by reading the text even though it promises a simple decentralized solution.

mnabozny 7y ago
Sure, here is simple hello world example: https://gitlab.cloudover.io/dinemic/starsheep-py/blob/master/examples/hello_world/hello_world.yaml https://gitlab.cloudover.io/dinemic/starsheep-py/blob/master... It creates simple model "Minion" and adds new listener, which is executed each time when field "name" is changed in any Minion instance. Listener is set to call bash script with "Hello [minion name]". Second part of yaml just creates three minions and sets theirs names. That triggers executing listener and script on each machine with this yaml. You can play with local copy of that file on two different machines and debug mode: sudo starsheep --launch --debug ./hello_world.yaml First machine will create that three original minions. Any attempt of modification of that minions, made on any other machine will be rejected on whole cluster as not authorized (not created by owner). Also you could play with changing minion names, to create new ones. Another example is here: https://gitlab.cloudover.io/dinemic/starsheep-py/tree/master/examples/mjpeg_streamer https://gitlab.cloudover.io/dinemic/starsheep-py/tree/master... It allows to automate deployment of mjpg_streamer on Raspberry PI, to share camera. It also shows how to split the application definition (models, listeners, scripts and variables) and data section. There should be available section for "motion" recorder soon
- eterps 7y ago
  Yes, this gives an example. But when would someone use this? Any use cases that are a good fit?
  - mnabozny 7y ago
    Initially it was designed to manage IT infrastructure and to automate server management even during "split brains", for each partition individually. It was intended to be partially a replacement for Ansible and partially as something more functional, with logics. Now, when I'm working a bit more with this tools, I think that it could be applied to anything related to IoT, autonomous cars/robots or wide area installations, like BTS network and its configuration or management of home/industrial automation. It should fit good in places, where connectivity is far away from perfect and we cannot guarantee connection with central management service (let's say database or LDAP), and regardless the network damages, all remaining services should be available and functional as much as possible. As far as your question is concerned, initially I planned to use that to manage virtual machines, bare metal servers and my local machines to provision them regardless the cloud operator, connection type and cloud automation services. For some reasons Ansible was too complicated to deliver the same functionality as I needed. Moreover, with Ansible there always be some "central" service that manages whole infrastructure. For client, who has some infrastructure in cloud and some in own, local servers, there would always be a weak point in managing such infrastructure. From the other hand there was a Kubernetes, but it was too complicated, especially when spanning over local cluster and public cloud
    - eterps 7y ago
      Thank you
- isusmelj 7y ago
  Thanks a lot, I'll have a look at the links you shared!

Could someone give me an example of when I would use it? It seems to me rather complicated by reading the text even though it promises a simple decentralized solution.

mnabozny 7y ago

Sure, here is simple hello world example: https://gitlab.cloudover.io/dinemic/starsheep-py/blob/master/examples/hello_world/hello_world.yaml https://gitlab.cloudover.io/dinemic/starsheep-py/blob/master... It creates simple model "Minion" and adds new listener, which is executed each time when field "name" is changed in any Minion instance. Listener is set to call bash script with "Hello [minion name]". Second part of yaml just creates three minions and sets theirs names. That triggers executing listener and script on each machine with this yaml. You can play with local copy of that file on two different machines and debug mode: sudo starsheep --launch --debug ./hello_world.yaml First machine will create that three original minions. Any attempt of modification of that minions, made on any other machine will be rejected on whole cluster as not authorized (not created by owner). Also you could play with changing minion names, to create new ones. Another example is here: https://gitlab.cloudover.io/dinemic/starsheep-py/tree/master/examples/mjpeg_streamer https://gitlab.cloudover.io/dinemic/starsheep-py/tree/master... It allows to automate deployment of mjpg_streamer on Raspberry PI, to share camera. It also shows how to split the application definition (models, listeners, scripts and variables) and data section. There should be available section for "motion" recorder soon

eterps 7y ago
Yes, this gives an example. But when would someone use this? Any use cases that are a good fit?
- mnabozny 7y ago
  Initially it was designed to manage IT infrastructure and to automate server management even during "split brains", for each partition individually. It was intended to be partially a replacement for Ansible and partially as something more functional, with logics. Now, when I'm working a bit more with this tools, I think that it could be applied to anything related to IoT, autonomous cars/robots or wide area installations, like BTS network and its configuration or management of home/industrial automation. It should fit good in places, where connectivity is far away from perfect and we cannot guarantee connection with central management service (let's say database or LDAP), and regardless the network damages, all remaining services should be available and functional as much as possible. As far as your question is concerned, initially I planned to use that to manage virtual machines, bare metal servers and my local machines to provision them regardless the cloud operator, connection type and cloud automation services. For some reasons Ansible was too complicated to deliver the same functionality as I needed. Moreover, with Ansible there always be some "central" service that manages whole infrastructure. For client, who has some infrastructure in cloud and some in own, local servers, there would always be a weak point in managing such infrastructure. From the other hand there was a Kubernetes, but it was too complicated, especially when spanning over local cluster and public cloud
  - eterps 7y ago
    Thank you
isusmelj 7y ago
Thanks a lot, I'll have a look at the links you shared!