On 3/14/07, Karel Gentens <[EMAIL PROTECTED]> wrote:
The zmd process i had shut down in the past using yast. that process does not run any more.
So you have Linux usage (reported by VM) that is not noticed by the RMFPMS agents or by top. It would be interesting to look how good your capture ratio is with some more load on the system because I don't think we're looking at a fixed 1-2% overhead. The approach used by RMFPMS is rather simplistic and by definition only works in some trivial cases. You will not have a high capture ratio in most real-life workloads. We even have cases where you would get less than 10% of the usage measured, even when burning a lot of resources on the measurement itself. Obviously you can make little conclusions on 10% of the data when you don't even know which 10%. Let alone doing charge-back on application or user. This is why ESALPS uses a different approach to measure process usage. Believe me, it's non-trivial to do this. And we do get high capture ratio (and at lost cost in most scenario's). Good enough to even do charge-back per application and per user. Rob -- Rob van der Heij Velocity Software, Inc http://velocitysoftware.com/ ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
