I have a SunRay server that I am looking at to determine some sizing requirements in my department. The machine has 16G of ram and 10G of swap. Currently, I have about 4G of swap used. I am wondering if dtrace/mdb can be used to find out what lwp/processes have been swapped out? Any hints? This message posted from opensolaris.org
Brendan Gregg - Sun Microsystems
2006-Oct-26 19:08 UTC
[dtrace-discuss] What has been swapped out?
G''Day Matthew, On Thu, Oct 26, 2006 at 10:14:02AM -0700, Matthew C Aycock wrote:> I have a SunRay server that I am looking at to determine some sizing > requirements in my department. The machine has 16G of ram and 10G of swap. > Currently, I have about 4G of swap used. I am wondering if dtrace/mdb can > be used to find out what lwp/processes have been swapped out?If paging or swapping to the physical swap device is causing an ongoing performance problem, then there are some things we can do with DTrace as events are happening for DTrace to measure. (prstat -m can also show processes being affected - look for increases in DFL). However if you want to walk up to a server and run a command to list physical swap device usage by process, for sizing requirements, then I doubt we can do much with DTrace. mdb is more likely to solve this by walking through anonymous address segments. procfs should probably provide this information (eg, /proc/<pid>/xmap), but currently doesn''t. no worries, Brendan -- Brendan [CA, USA]
Following link may be useful to you, http://prefetch.net/articles/dtracecookbook.html Look for session "Anonymous pages paged in (api)". -ey Brendan Gregg - Sun Microsystems wrote:> G''Day Matthew, > > On Thu, Oct 26, 2006 at 10:14:02AM -0700, Matthew C Aycock wrote: > >> I have a SunRay server that I am looking at to determine some sizing >> requirements in my department. The machine has 16G of ram and 10G of swap. >> Currently, I have about 4G of swap used. I am wondering if dtrace/mdb can >> be used to find out what lwp/processes have been swapped out? >> > > If paging or swapping to the physical swap device is causing an ongoing > performance problem, then there are some things we can do with DTrace > as events are happening for DTrace to measure. (prstat -m can also > show processes being affected - look for increases in DFL). > > However if you want to walk up to a server and run a command to list > physical swap device usage by process, for sizing requirements, then > I doubt we can do much with DTrace. mdb is more likely to solve this > by walking through anonymous address segments. procfs should probably > provide this information (eg, /proc/<pid>/xmap), but currently doesn''t. > > no worries, > > Brendan > >-- Erhyuan (EY) Tsai Accessline: X40209/650-472-1468 Performance Engineer PAE/Performance Availabilty, and Architecture Engineering -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20061026/ca606921/attachment.html>