1 |
2011/3/1 Peter Humphrey <peter@××××××××××××××.org>: |
2 |
> On Tuesday 01 March 2011 23:14:12 Mick wrote: |
3 |
> |
4 |
>> Ha! I remember on an old machine when in WinXP would rarely if ever |
5 |
>> crash, while in Gentoo would crash every time. |
6 |
> |
7 |
> My machine is only about a year old, built by a specialist builder of high- |
8 |
> performance systems, so it shouldn't be experiencing hardware failures. |
9 |
> |
10 |
>> Different OS' use memory differently. |
11 |
> |
12 |
> Indeed they do. My experience is the converse of yours: Gentoo does not |
13 |
> hang, while Fedora and Mandriva do. It's not a problem with a particular |
14 |
> area of the disks, as I've installed them both in different partitions and |
15 |
> got the same result. I assume that some kernel options don't suit my |
16 |
> motherboard. Don't all laugh, but it's an Asus P7P55D. |
17 |
> |
18 |
>> After a year or so though the WinXP installation eventually corrupted |
19 |
>> itself irreparably, while Gentoo (on reiserfs) soldiered on. Eventually, I |
20 |
>> bought new memory modules and there were no more crashes. |
21 |
> |
22 |
> Maybe I need to replace the memory. That's a bit drastic though when I |
23 |
> haven't actually proved it faulty. |
24 |
|
25 |
Yes, I tend to agree. You could end up replacing the memory only to |
26 |
find out that the crashes persist. |
27 |
|
28 |
|
29 |
>> memtest 86+ showed no errors, so I didn't know what to blame for all |
30 |
>> these crashes. |
31 |
> |
32 |
> It's well known that test programs can't stress a computer the way real life |
33 |
> does. It was true of Ferranti Argus 500 systems in 1974, and I'm sure it's |
34 |
> still true today. |
35 |
|
36 |
I remember using a script which put the system (memory modules and |
37 |
swap) through its paces. That did show me some errors which made me |
38 |
replace the memory. I can't recall where I found that script, but |
39 |
remember it being aired in this mailing list. |
40 |
|
41 |
|
42 |
>> After close observation I discovered that the machine would crash the |
43 |
>> moment it tried to start swapping. |
44 |
> |
45 |
> Interesting. As far as I can tell though this box doesn't swap often - it |
46 |
> can go weeks without doing so. As I said the other day, my 4GB is enough to |
47 |
> contain the work I usually do. |
48 |
> |
49 |
>> This would typically happen in the middle of an emerge, which was rather |
50 |
>> annoying, and/or when updatedb was running. |
51 |
> |
52 |
> At least you could re-run an aborted emerge; when my box hangs it just stops |
53 |
> responding to keyboard and mouse, and the network interface stops receiving |
54 |
> packets so I can't ssh in from another box to shut it down neatly. It's BRS |
55 |
> time. |
56 |
|
57 |
No I couldn't. :-( |
58 |
|
59 |
The whole system would freeze up, no keyboard, no network, no nothing. |
60 |
I had to pull the plug every time. |
61 |
-- |
62 |
Regards, |
63 |
Mick |