On Wednesday 12 November 2003 02:24 am, Damian Gerow wrote:
| Thus spake "Brian T. Schellenberger" <[EMAIL PROTECTED]> [01:57:43 11/12/03:
| : Yeah, but the browser is irrelalvent; 'tis fproxy that handles the
| : freenet stuff for the browswer, and *it* knows about freenet.
| :
| : Not that it's likely to do better the fuqid . . . but it might be
| : interesting to give it a go.
|
| Take a look at ulimit, and the following options as defined in
| sys/i386/conf/LINT:
|
| MAXDSIZ="(256*1024*1024)"
| MAXSSIZ="(256*1024*1024)"
| DFLDSIZ="(256*1024*1024)"
|
| In more than a couple of cases, I've upped this a bit higher. There's some
| notes right around that part of LINT that explain what each does.
|
| You're probably most interested in increasing MAXDSIZ and MAXSSIZ, to what
| depends on what else the machine is doing, and how much RAM you have
| available.
|
| (FWIW, to get the most out of a MySQL install, I've had these two as high
| as 1024*1024*1024 -- so a GB, as opposed to 256MB by default.)
Hmm . . . something odd would seem to be going on, since I don't specify these
at all in my kernel, yet I have seen the freenet java process in top showing
as consuming well over 1G (at the moment it's 823M), and changing the java
args to -Xmx512M made a very noticable difference in how well freenet ran.
But if the kernel truly defaulted to 128M per-process limit, then surely it
would make no difference if I told java to use 0.5G, nor could it be
consuming 1.2G (the highest I've seen yet) total.
What am I missing here?
--
Brian, the man from Babble-On . . . . [EMAIL PROTECTED] (personal)
_______________________________________________
Support mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
