1 |
On Fri, 22 Jul 2005 17:52:09 -0600 |
2 |
Joseph <syscon@×××××××××.com> wrote: |
3 |
|
4 |
|
5 |
> > >Does anybody has any other solutions? |
6 |
> > >i |
7 |
|
8 |
There are a few tools that will allow you to do some diagnosing. |
9 |
|
10 |
These will isolate your harddrive and drive controllers. |
11 |
|
12 |
app-benchmarks/bonnie (2.0.6): Performance Test of Filesystem I/O using standard C library calls. |
13 |
app-benchmarks/bonnie++ (1.93c): Hard drive bottleneck testing benchmark suite. |
14 |
|
15 |
If is is the motherboard, it should fall over pretty quick. |
16 |
|
17 |
Another tool I like is - |
18 |
|
19 |
app-benchmarks/stress (0.18.6): Imposes stressful loads on different aspects of the system. |
20 |
|
21 |
You'll have to add - app-benchmarks/stress x86, to your /etc/portage/package.keywords |
22 |
as they don't have the amd64 keyword in the ebuild. It builds and runs fine. |
23 |
|
24 |
Stress allows you to load all or parts of the system up for a defined period of time. It's |
25 |
even possible to run the system out of resources. It's a real nice test of system stabilty. |
26 |
All except the Xserver and that's easy to add by running 3 of the rss-glx screensavers |
27 |
from a term while running stress. And if you make the virtual memory component large |
28 |
enough at runtime, the system will start swapping. |
29 |
|
30 |
This line will get the load up to about 20 and cause about 500MB of swapping |
31 |
to occur on a 1P amd64 system with 1 GB of main memory - |
32 |
|
33 |
stress --cpu 16 --io 4 --vm 2 --vm-bytes 1024M --timeout 60s -d 2 |
34 |
|
35 |
Change the timeout to be around 5 minutes or 600 seconds. Get a tail -f /var/log/messages |
36 |
or use root-tail. And get a top running in another term. |
37 |
|
38 |
Bob |
39 |
- |
40 |
-- |
41 |
gentoo-user@g.o mailing list |