IBM Mainframe Discussion List <[email protected]> wrote on 07/15/2006 06:18:31 PM:
> Speaking of delays to get an SRB going, why are there sometimes such HUGE > delays? I have seen it take upwards of 1/10 of a second to dispatcha global > SRB on a 48 MIPS (whatever that is) CPU. And since cross-memory post > schedules an SRB to do the post, that means a cross-memory post may > also take upwards of 1/10 of a second. That's kind of hard to answer without a system trace. If you have programmed an SRB routine to determine when it has taken a long time to be dispatched, have it initiate an SDUMP in order to capture system trace. TRAPS NAME(IEASCHEDULETRACE) in DIAGxx will cause the SCHEDULE to be traced in system trace. TRAPS NAME(IEASCHEDULETRACE,IEANOSUSPSYSTRC) may be better for this particular issue, since it will also trace CPU timer and clock comparator interrupts out of wait, which we normally do not trace in order to avoid wrapping more useful entries out of the trace. Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY ---------------------------------------------------------------------- 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
