Agreed. I missed that. Since it does only the process,
is that good enough?
Jim
----
Chad Mynhier wrote:
On Sun, Jul 12, 2009 at 1:28 PM, James
Litchfield<james.litchfi...@sun.com> wrote:
It goes back to what problem is being solved.
A) ptime requires that it start the app. It can't
attach to a running app. It aggregates all the
data and does not offer a per-lwp breakdown.
Given the mechanisms it uses, that's all it can do.
If that's adequate then the problem is solved.
Actually, the version of ptime that Jim's talking about _can_ attach
to a running app. I added the -p option to do just that. (Of course,
it won't attach to a running app and then wait until it completes, it
only attaches long enough to grab a snapshot of the microstate stats.)
For example:
# ptime -mp `pgrep nscd`
real 29:45:57.427092032
user 1.852038785
sys 2.133489810
trap 0.005027352
tflt 0.000000000
dflt 0.000009597
kflt 0.000000000
lock 29:45:54.115500466
slp 773:53:38.737153459
lat 0.105180668
stop 0.017284688
#
Chad
_______________________________________________
dtrace-discuss mailing list
dtrace-discuss@opensolaris.org
