Why are no processes swapped in FreeBSD-1.0?
Warren Toomey
are ever swapped out to disk (i.e there are no IW processes marked). However,
under 386bsd-0.1+patchkits, this does happen. I've ensured that the system has
had enough time and inclination to swap out some processes (e.g processes
idle for >30 seconds, lots of memory active in other processes).
Can somebody tell me if this is a problem in FreeBSD-1.0, or if ps itself
is not properly reporting process status.
P.S Why doesn't iostat show disk activity :-) ???
Many thanks!
Warren
--
`` ... nor do I regularly post multi-kilobyte screeds about driver software I
can't release because aliens haven't used the radio receivers in my head to
give me permission yet.'' -- John F. Woods
Nate Williams
Warren Toomey <w...@cserve.cs.adfa.oz.au> wrote:
>Looking at the output of ps -aux under FreeBSD-1.0, it seems that NO processes
>are ever swapped out to disk (i.e there are no IW processes marked). However,
>under 386bsd-0.1+patchkits, this does happen. I've ensured that the system has
>had enough time and inclination to swap out some processes (e.g processes
>idle for >30 seconds, lots of memory active in other processes).
From what I understand given the original VM system (I'm not expert, but
one of our VM experts is not on Internet and the other one is too busy)
that 386BSD did not swap like typical unix systems. Maybe 386BSD did not
print out the correct statistics? (See notes below)
>Can somebody tell me if this is a problem in FreeBSD-1.0, or if ps itself
>is not properly reporting process status.
If it's any consolation, FreeBSD 1.1 does indeed swap.
....
nate 83 0.0 0.0 52 0 v0 IW+ 5:49PM 0:00.28 xinit
root 76 0.0 0.0 24 0 v1 IWs+ 5:48PM 0:00.08 /usr/libexec/ge
nate 75 0.0 0.0 412 0 v0 IWs 5:48PM 0:00.96 -tcsh (tcsh)
Here's a bit out of some notes John posted to the mailing lists.
...
As you might know, the version of the MACH VM system that
is being used by most of the BSD variants is not really complete
and has some problems. The inadequacies included:
..
3) Non-implementation of swapping.
4) Paging in/out of ONLY single pages.
..
9) The system will hang on swap space full.
But FreeBSD now has..
4) True swapping support, tied in with RSS limits and UPAGE
paging. The UPAGES have been moved into process space, and
they are paged when a process is swapped out. (UPAGES MUST
be wired when a process is swapped in.)
5) The page daemons have been essentially redone, and vm_fault
has had significant additions and improvements to support
CLUSTER paging.
..
16) The system can relieve its problem of running out of
swap space, eliminating an impending hang.
---------------------------
As you can, I only posted a few of the problems and solutions that the
new VM system attempts to solve since those are the problems spoken
about above, but there are many more.
>P.S Why doesn't iostat show disk activity :-) ???
Because no-one has (yet) integrated the counters into the disk drivers.
(There's a project for someone to do. :-)
Nate
--
na...@bsd.coe.montana.edu | FreeBSD core member and all around tech
na...@cs.montana.edu | weenie.
work #: (406) 994-4836 | Graduating May '94 with a BS in EE
home #: (406) 586-0579 | - looking for work in CS/EE field.
Garrett Wollman
>are ever swapped out to disk (i.e there are no IW processes marked). However,
>under 386bsd-0.1+patchkits, this does happen. I've ensured that the system has
>had enough time and inclination to swap out some processes (e.g processes
>idle for >30 seconds, lots of memory active in other processes).
The swapper in 386BSD never worked properly. (It would mark the
process as swapped, but would never actually swap anything out.) For
FreeBSD 1.0, we disabled it because it actually resulted in worse
performance. In FreeBSD 1.1, large portions of the VM system have
been completely re-written, and real swapping is actually implemented.
The VM system is also an order of magnitude or more faster than in
1.0.2.
So, in other words, FITNR.
>P.S Why doesn't iostat show disk activity :-) ???
Because the disk stats were never implemented. This is on the TODO
list prety immediately following 1.1. (We're in a feature freeze at
the moment, so it won't happen in time for 1.1.)
-GAWollman
--
Garrett A. Wollman | Shashish is simple, it's discreet, it's brief. ...
wol...@emba.uvm.edu | Shashish is the bonding of hearts in spite of distance.
NB: Address soon to | It is a bond more powerful than absence. We like people
be changed. | who like Shashish. - Claude McKenzie + Florent Vollant