I don't understand your rss comment. I thought that setting the limit field
effectively gets you a rss limit, no?!

> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] Behalf Of Chandra
> Seetharaman
> Sent: Wednesday, October 13, 2004 1:07 PM
> Cc: [EMAIL PROTECTED]
> Subject: Re: [Fwd: Re: [ckrm-tech] memory resource controller for e16]
>
>
> On Tue, Oct 12, 2004 at 12:30:06AM -0500, Ray Bryant wrote:
> Ray,
>
> Thanks for your comments..
>
> We are not distinguishing _between_ active and inactive. It is the sum of
> active and inactive pages that is used against limit/guarantee.
>
> Agreed, rss can also be controlled, we thought if needed that can be
> implemented as a different resource controller....
>
> chandra
> >
> > Chandra,
> >
> > Sorry for a late post on this, but I just happened to be
> perusing this list,
> > and I was curious why the CKRM memory resource controller cares about
> > distinguishing between active and inactive pages.
> >
> > It seems to me that the "share" that should be guarenteed to a
> program is
> > its
> > use of main memory (e. g. mm->rss) and not of a certain number
> of pages on
> > the
> > active or inactive list.  The resource that is being controlled is main
> > memory
> > pages, not slots on the lru lists.
> >
> > Indeed, one can readily construct cases where the Linux vmm
> will put all of
> > the pages for a particular application on the active list (or for that
> > matter,
> > the inactive list) while the total number of pages used is
> precisely the
> > same.
> >
> > Examples of this abound, but here is a simple one to try:  set
> swappiness to
> > zero, malloc and touch some memory in a process, sleep and repeat the
> > touches;
> > then start up a big I/O (file copy) in the background that will touch
> > sufficient pages to force memory to be overcomitted.  The vmm
> will put all
> > of
> > the malloc processs's pages onto the active list.  This process has done
> > nothing differently after the big I/O process has started than it did
> > before,
> > and its pages will get moved.
> >
> > If you artificially set up a small active list share (smaller than the
> > number
> > of pages required by this process), how will the controller
> deal with this,
> > kill the process?  (It's not the process's fault.....)
> >
> > At least the way it seems to me is that its not going to be
> productive to
> > specify process level memory shares in terms of these lists.  A more
> > reasonable set of shares, it would seem to me, would be
> limits/guarentees on
> > the amount of ZONE_NORMAL, ZONE_DMA, and ZONE_HIGH memory used
> by a process.
> >
> > Now this is all based on a quick perusal of your design
> document.  If I've
> > misread or misunderstood this in some gross way, my sincerest apologies.
> >
> > --
> > Best Regards,
> > Ray
> > -----------------------------------------------
> >                   Ray Bryant
> > 512-453-9679 (work)         512-507-7807 (cell)
> > [EMAIL PROTECTED]             [EMAIL PROTECTED]
> > The box said: "Requires Windows 98 or better",
> >            so I installed Linux.
> > -----------------------------------------------
> >
> >
> >
> > --
> > Best Regards,
> > Ray
> > -----------------------------------------------
> >                   Ray Bryant
> > 512-453-9679 (work)         512-507-7807 (cell)
> > [EMAIL PROTECTED]             [EMAIL PROTECTED]
> > The box said: "Requires Windows 98 or better",
> >            so I installed Linux.
> > -----------------------------------------------
> >
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> > Use IT products in your business? Tell us what you think of
> them. Give us
> > Your Opinions, Get Free ThinkGeek Gift Certificates! Click to
> find out more
> > http://productguide.itmanagersjournal.com/guidepromo.tmpl
> > _______________________________________________
> > ckrm-tech mailing list
> > https://lists.sourceforge.net/lists/listinfo/ckrm-tech
>
> --
>
> ----------------------------------------------------------------------
>     Chandra Seetharaman               | Be careful what you choose....
>               - [EMAIL PROTECTED]   |      .......you may get it.
> ----------------------------------------------------------------------
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> Use IT products in your business? Tell us what you think of them. Give us
> Your Opinions, Get Free ThinkGeek Gift Certificates! Click to
> find out more
> http://productguide.itmanagersjournal.com/guidepromo.tmpl
> _______________________________________________
> ckrm-tech mailing list
> https://lists.sourceforge.net/lists/listinfo/ckrm-tech



-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
ckrm-tech mailing list
https://lists.sourceforge.net/lists/listinfo/ckrm-tech

Reply via email to