Raspberry distributed computing
Greetings to all,
This is the first operative post of our new blog. You may wondering what we are doing after the release of Muster 9 that required 1.5 years of development time.
At the moment, we are polishing the build of Muster 9 for the Raspberry PI 3 platform. Several customers asked us about the possibility to have the clients components of Muster 9 on a Rasp to handle repetitive low cpu demanding tasks on a cheap platform.
The good news is that Muster 9 is the only distributed computing solution available on the market built in standard POSIX C++ and uses native Linux APIs and platforms facilities without using external big frameworks like the .NET platform or Mono. We have built it successfully for the Raspberry PI 3 using the default Raspbian distribution and it will be available next week within the first Muster service release 9.0.1 that will also include some minor enhancements (no bugs fix until now, phew!).
I’ve to admit that this tiny board surprised me. While I’ve played with it in the past to implement weird home automation stuff, suggestion to implement it inside Muster is not our idea but has been picked up directly from our customers ones. Muster 9 works very well on it allowing raspberry distributed computing as a Renderclient. You can even use it as a low cost Dispatcher service to handle a quite big amount of Renderclients, or as a cron based system to interact with Muster with mrtool or the Python APIs doing all sort of things.
We are also working on several improvements on the chunks and jobs view that will let you browse jobs and chunks statistics like cpu usage over time and overall cpu and RAM usage, plus some interesting additional tools. After the release of this version, we’ll keep you informed of the Muster 9 2018 roadmap.