Re: 2.2.14 feels slow and non-interactive.

From: Jorgen Skjaanes (js@pimp.gulesider.no)
Date: Wed Jan 19 2000 - 21:34:07 EST


On Wed, 19 Jan 2000, Mike A. Harris wrote:

> I've been using 2.2.14 since it came out, and it has had a
> sluggish feel to it at times. When switching from X to the
> console and back often I have to wait anywhere up to 5-10 seconds
> for my mouse to move, or to be able to type, or for the screen to
> stabilize. In netscape, if I'm clicking on a pulldown or
> bookmark folder on the personal toolbar folder, I have to wait
> up to 5-10 seconds as well for it to open up. It seems as if it
> is pulling it out of swap space or something. X starts up slow
> as well, and when PINE checkpoints my 20Mb email folders it takes
> much longer or so it seems.
>
> If I switch back to an identically configured 2.2.13 or 2.2.12
> the "problem" goes away and things seem smooth again. I don't
> know if there were scheduler changes or not in 2.2.14, but it
> sure doesn't seem very interactive.
>
> My system is a K6-200 with 96 Megs of RAM. The kernels that I'm
> using are stock Official kernels from ftp.kernel.org with NO
> external patches and NO 3rd party hacks or kernel modules. Just
> a plain Jane kernel.
>
> Anyone else experience loss of interactivity and slowdowns on
> 2.2.14?

Well, yes, I think so. The problem is I don't have a good "before"
situation to compare with. I installed a database service doing
some logging every 5 minutes and upgraded from 2.2.8-pre5 at the
same time. After that when the database is doing its stuff both
the database process and netscape gets stuck in D-state in wait_on_page
for many seconds while the disk (Adaptec AIC7xxx contr) is trashing
like mad. If I close netscape during this I can see the netscape
process staying in wait_on_page for up to 15 seconds before it exits.
The computer (a PPro-200, 160MB) is being so unresponsive that I've
ordered a new one. I sort of thought it was just being overloaded,
and maybe it is, but processes being stuck in D-state for such a long
time doesn't seem right. And since you asked...

And it's not swap related. Or at least vmstat doesn't say so. Though
it is allways deep into swap. Right now when X is not running:

             total used free shared buffers cached
Mem: 160512 157532 2980 39880 3276 115928
-/+ buffers/cache: 38328 122184
Swap: 72288 26980 45308

( I know this is not a good and detailed report, but since I haven't seen
  anyone else seeing this I still am not sure it's a bug. And I don't want
  to flood the list with ps output and every litle detail if it isn't.)

-- 
Jorgen

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Jan 23 2000 - 21:00:21 EST