| 1 |
[snip] |
| 2 |
|
| 3 |
> I realize processes can be swapped in and out as necessary. But I did not |
| 4 |
> expect a server coming up under no load to display this. I don't see this |
| 5 |
> behavior on my Redhat severs. This sever has 512 MB of RAM. top shows |
| 6 |
> about 62 MB of RAM used and all but one process is sleeping. |
| 7 |
> |
| 8 |
> Using 2.4.20-gentoo-r7 kernel. |
| 9 |
> |
| 10 |
> Is this normal behavior for Gentoo? Or did I miss something in setup? |
| 11 |
|
| 12 |
Try using a saner kernel. I'm not the biggest fan of the present |
| 13 |
gentoo-sources. It's based on an old -ck patch for an aging branch, and my |
| 14 |
first suspicion is that some of Kolivas' hacks could be reponsible for the |
| 15 |
effect. |
| 16 |
|
| 17 |
I use 2.4.20-wolk4.10-pre7 on my server (currently wolk-sources-4.9 is in |
| 18 |
Portage) and do not experience any such behaviour. The gs-sources kernel |
| 19 |
is pretty good (if you don't mind running a 2.4.23-pre) but features no |
| 20 |
O(1) scheduler and the patchset is quite conservative, albeit balanced. |
| 21 |
pfeifer-sources is the testbed for the next gentoo-sources I think, and I |
| 22 |
believe its name will be changed shortly to reflect that. |
| 23 |
|
| 24 |
But really, I would consider using wolk-sources, gs-sources-2.4.23_pre7 or |
| 25 |
aa-sources-2.4.22-r1). ac-sources is probably a good kernel too, but I |
| 26 |
haven't much experience there. |
| 27 |
|
| 28 |
HTH, |
| 29 |
|
| 30 |
--Kerin |
Archives