PipelineFX Forum

Qube! => General => Topic started by: methodhai on March 07, 2008, 09:44:59 PM

Title: Explanation of groups and cluster
Post by: methodhai on March 07, 2008, 09:44:59 PM

Hi,

I'm working on some documentation for our users and it occurred to me that I couldn't easily explain the difference between groups and cluster since I lacked the details of how they operate - and how they fit into the scheme of Qube. After some digging in the docs - it seems that groups are permeable and clusters are not. But I couldn't find in the docs a clear explanation of how the two play together - might have just missed it.

I was wondering if I could get a more detailed explanation the Qube groups/clustering concepts.

Thanks,
Hai

Title: Re: Explanation of groups and cluster
Post by: eric on March 10, 2008, 11:51:46 PM

Quote from: methodhai on March 07, 2008, 09:44:59 PM

Hi,

I'm working on some documentation for our users and it occurred to me that I couldn't easily explain the difference between groups and cluster since I lacked the details of how they operate - and how they fit into the scheme of Qube. After some digging in the docs - it seems that groups are permeable and clusters are not. But I couldn't find in the docs a clear explanation of how the two play together - might have just missed it.

I was wondering if I could get a more detailed explanation the Qube groups/clustering concepts.

Thanks,
Hai

You have that backward. Groups are non-permeable, while clusters are. A group can be thought of as an alias to a list of hosts. The job can only run on the list of hosts or host group. If nothing is available, the job will pend.

A cluster is a priority scheme where jobs targeted to a specific cluster will have higher priority over jobs with submitted to a less specific cluster.

So if you have a host in cluster "/clusterA," and a 2 jobs, jobA submitted to "/" and jobB to "/clusterA," jobB will have a higher priority over jobA, on that host.