OpenVZ allows for containers to be provided with "burst" memory and CPU allocations. When this style configuration is implemented on a container, /proc/cpuinfo and /proc/meminfo return this allocation, which really isn't accurate. Right now we have to include a good chunk of logic in our recipes to determine the proper values.
Instead, if the system is discovered as an OpenVZ guest (See http://tickets.opscode.com/browse/OHAI-249), we should parse /proc/user_beancounters and determine the "guaranteed" memory & CPU allocations.
I'm not entirely sure if this code should be placed here:
Or if we should place the code here:
And have it override any values discovered by cpu.rb & memory.rb.
Any insight/suggestions would be appreciated! Thanks!
|Field||Original Value||New Value|
|Workflow||jira [ 12938 ]||New OSS [ 22791 ]|
|Status||Open [ 1 ]||Confirmed [ 10001 ]|
|Status||Confirmed [ 10001 ]||Fix Provided [ 10002 ]|
|Status||Fix Provided [ 10002 ]||Reopened [ 4 ]|
|Workflow||New OSS [ 22791 ]||OSS: Product Management [ 28797 ]|
|Status||Reopened [ 4 ]||Open [ 1 ]|
|Transition||Time In Source Status||Execution Times||Last Executer||Last Execution Date|
|665d 8h 32m||1||Chris Lundquist||09/Jan/13 12:38 AM|
|2s||1||Chris Lundquist||09/Jan/13 12:38 AM|
|47s||1||Chris Lundquist||09/Jan/13 12:39 AM|
|62d 20m||1||Bryan McLellan [Opscode]||12/Mar/13 12:59 AM|