I'd be interested to see what Dave's capture ratio is? In our case the problem manifested itself, on a small 2-way, by one of our WAS V5 servants using excessive CPU along with a dramatic drain of CPU, to dispatchable work, by a uni-swap loop which was basically being done by 3 swappable WAS servants playing musical chairs on the LPAR. Our capture ratio during these times was probably 50 - 60% with 1k - 2k/swaps a minute. Turned out to be an SRM defect with express swap logic as defined by IBM. Not a pretty picture - having everyone sitting on their hands during the problems.
Shane Ginnane <[EMAIL PROTECTED]> Sent by: IBM Mainframe Discussion List <[email protected]> 09/27/2005 07:22 PM Please respond to IBM Mainframe Discussion List <[email protected]> To [email protected] cc Subject Re: z/OS 1.4 Slowdowns "The most common causes of uncaptured system overhead are for excessive paging and a internal queue build up of SRBs (usually for cross-memory POST) due to server address spaces having a priority lower than their clients. The later is impossible to see with monitors." Shane ... ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
