Update: Increasing the virtual memory size of the PERFSVM userid solved the problem. I wish PerfKit gave more useful messages.... I¹ve been chasing the wrong problem all day.
-- .~. Robert P. Nix Mayo Foundation /V\ RO-OE-5-55 200 First Street SW / ( ) \ 507-284-0844 Rochester, MN 55905 ^^-^^ ----- "In theory, theory and practice are the same, but ³Join the story... Ride Ural.² in practice, theory and practice are different." On 7/11/07 1:12 PM, "Kris Buelens" <[EMAIL PROTECTED]> wrote: > This is not related the size of a minidisk, but the size of a DisContiguous > Saved Segment, a piece of virtual storage, in this case shared between CP > (that stores the performance data in it ) and PERFSVM that retrieves them out > of it. > > It is created by issuing DEFSEG MONDCSS sss-eee SC (I'm a bit uncertain > about "SC") > followed by SAVESEG MONDCSS > sss end eee are the hex starting page address and ending address > You can issue Q NSS NAME MONDCSS MAP to find its current size. > To have the new version active, all users of it should be stopped. Q NSS > USERS MONDCSS will show the current users, CP itself will not be listed, but > if you issue MONITOR STOP, CP will stop using it too. > > 2007/7/11, RPN01 <[EMAIL PROTECTED]>: >> We've been running PerfKit for quite some time, and the 191 disk got too >> small for the trend data. I brought down both PERFSVM virtual machines (we >> run two CECs and CSE), created new, larger disks, and set up the PERFSVM's to >> use them, and then xautolog'ed both PERFSVMs. >> >> Now, both are getting the message "FCXPMN446E Incomplete monitor data: DCSS >> size too small". >> >> The current data was just copied from one disk to another; what does the DCSS >> size have to do with that? If the DCSS is too small, what is the yardstick >> needed to decide how big to make it?