(I understand this is somewhat tangential to the original email, I will just clarify my thought in previous email. That said, this is not a fully baked idea, just thinking out loud.)
"init" is the solution, right. However, when dealing with 1000s of systems, one would need automation for managing init on all of them. This isn't new, solutions exist. And I don't mean to suggest we run tasks via this mechanism, but services that would otherwise be configured via init, for example. In the future, I can imagine one would need "init" or the likes to work together with Mesos. Whether it is built into Mesos (the literal interpretation of what I suggested) or well integrated with other solutions with a clean interface is a design/implementation choice. Here's one possible scenario: A DataCenter runs Databases, Webservers, MicroServices, Hadoop or other batch jobs, stream processing jobs, etc. There's 1000s, if not 100s, of systems available for all of this. Ideally, systems running Databases are configured to run different services in their init than one running batch jobs. However, because one would want to achieve elasticity of different services (#systems running DBs vs. batch, for example) within a single Mesos cluster, Mesos would have to determine what services run on the system at the current time. It's like a newly installed system comes up and connects into Mesos and says, "hello there, I am an 8-core 48GB 1Gb eth system ready for service, what would you like me to do?". Mesos can choose to make it run any one or more of the services which would determine the set of init services to launch. And that may change over time as DC traffic changes. On Thu, Jan 22, 2015 at 11:56 PM, Ryan Thomas <[email protected]> wrote: > If this was going to be used to allocate tasks outside of the schedulers > resource management, and for every slave, why not just use the OS provided > init system instead? > > On 22 January 2015 at 19:40, Sharma Podila <[email protected]> wrote: > >> Schedulers can only use resources on slaves that are unused by and >> unallocated to other schedulers. Therefore, schedulers cannot achieve this >> unless you "reserve" slots on every slave for the scheduler. Seems kind of >> a forced fit. An init like support would be more fundamental to Mesos >> cluster itself, if available. >> >> >> On Thu, Jan 22, 2015 at 10:08 AM, Ryan Thomas <[email protected]> >> wrote: >> >>> This seems more like the responsibility of the scheduler that is >>> running, like marathon or aurora. >>> >>> I haven't tried it but I would imagine if you had 10 slaves and started >>> a job with 11 tasks with host exclusivity when you spin up an 11th slave >>> marathon would start it there. >>> >>> >>> On Thursday, 22 January 2015, Sharma Podila <[email protected]> wrote: >>> >>>> Just a thought looking forward... >>>> Might be useful to define an init kind of feature in Mesos slaves. >>>> Configuration can be defined in Mesos master that lists services that must >>>> be run on all slaves. When slaves register, they get the list of services >>>> to run all the time. Updates to the configuration can be dynamically >>>> reflected on all slaves and therefore this ensures that all slaves run the >>>> required services. Sophistication can be put in place to have different set >>>> of services for different "types" of slaves (by resource types/quantity, >>>> etc.). >>>> Such a feature bodes well with Mesos being the DataCenter OS/Kernel. >>>> >>>> >>>> On Thu, Jan 22, 2015 at 9:43 AM, CCAAT <[email protected]> wrote: >>>> >>>>> On 01/21/2015 11:10 PM, Shuai Lin wrote: >>>>> >>>>> OK, I'll take a look at the debian package. >>>>> >>>>> thanks, >>>>> James >>>>> >>>>> >>>>> >>>>> >>>>> You can always write the init wrapper scripts for marathon. There is >>>>>> an >>>>>> official debian package, which you can find in mesos's apt repo. >>>>>> >>>>>> On Thu, Jan 22, 2015 at 4:20 AM, CCAAT <[email protected] >>>>>> <mailto:[email protected]>> wrote: >>>>>> >>>>>> Hello all, >>>>>> >>>>>> I was reading about Marathon: "Marathon scheduler processes were >>>>>> started outside of Mesos using init, upstart, or a similar tool" >>>>>> [1] >>>>>> >>>>>> This means >>>>>> >>>>>> So my related questions are" >>>>>> >>>>>> Does Marathon work with mesos + Openrc as the init system? >>>>>> >>>>>> Are there any other frameworks that work with Mesos + Openrc? >>>>>> >>>>>> >>>>>> James >>>>>> >>>>>> >>>>>> >>>>>> [1] http://mesosphere.github.io/__marathon/ >>>>>> <http://mesosphere.github.io/marathon/> >>>>>> >>>>>> >>>>>> >>>>> >>>> >> >
