As much as I like the idea of having accounting data for each process/job in a linux server, I don't think the VM accounting data is the place to put it. For a long time I have felt that process accounting should go into the Monitor data stream so like SMF there is all the data in ONE place. Being a different beastie (like MVS or VSE), linux data is probably best kept in linux for a while with consolidation/coordination with ALL other sources of process accounting data being a future goal.
My current need is to get total CPU utilization per day and daily minutes of CPU utilization for certain subsets. But I have to go back in time for the data so my current PerfTK data is insufficient but the accounting cards cover the time period. For future rounds of this processing, using PerfTK data would be more complete but would be harder to process than the simple accounting cards. Besides once I start using accounting cards, I should continue with accounting cards. Any REXX code (no compilers available on this z/VM system) would be appreciated. /Tom Kern Scott Rohling wrote:
I mucked around with the source of a piece of the 'cpint' package once (thanks Neale!) -- to cut user (C1) accounting records from Linux.. It would get executed before a user 'job' was run on Linux, and after - to record the amount of usage. The account code to use was passed as part of the job. Unfortunately I don't have access to this anymore - but I'm not sure it's what you're really looking for so much as something to process accounting records? If so - I have some simple REXX code that does tallies of CPU time (ttime) from accounting records by userid.. meant to run on z/VM. Again - not sure this is what you're after.. Scott Rohling
---------------------------------------------------------------------- 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
