0

I hava scenario like this:

I have applications A,B,C,D..., and I hava physical machines M,N,O,P,Q...

I use byon to manage physical machine, because the physial machine is "strong", so I want to deploy several application on it, so I set the SLA is global, at this time I have a question: when application A is deployed on machine M, I deploy other application B,C,D...,whether application A,B,C,D...will install on M machine only, rather than install on machine N,O,P,Q...(in this case, the host A's pressure will be very large.) Is this problem exist, if exists, how to resolve it? thank you very much!

yzet
  • 11
  • 3

1 Answers1

1

It's possible to limit the number of services on a specific machine by specifying the memory required for each service. As part of the global isolation SLA You can set the amount of memory required by each service, so when there isn't enough memory left on the machine - the next one will be used.

The syntax is:

isolationSLA {      
    global { 
        instanceCpuCores 0
        instanceMemoryMB 128 // each instance needs 128MB allocated for it on the VM.
        useManagement true // Enables installing services on the management server. Defaults to false.
    }

Please note that the above code also allows services to be installed on the management machine itself, which you can set to false.

A more detailed explanation is available here, under "Isolation SLA".

Noa Kuperberg
  • 398
  • 2
  • 6
  • I set this SLA in jetty-service.groovy isolationSLA { global { instanceCpuCores 0 instanceMemoryMB 256 useManagement false } } But when I deploy an application with 2 instance, then two instance be install in same machine, at last, the result is one instance start successful, but another start failed, the failed one get an exception: "no management space located". I find this exception in org.cloudifysource.utilitydomain.context.kvstore.AttributesFacadeImpl.getManagementSpace() the exception in the pic: http://i.stack.imgur.com/9MVF9.png – yzet Jan 06 '15 at 15:34