On Tuesday 25 May 2010 09:28:29 starg...@gmx.sg wrote:
> I've finally decided to move from 0.5 to 0.7. I don't know if it's any of the 
> settings or the 0.7's hardware requirements - but the CPU is about 10 degrees 
> C warmer in average than before, and CPU usage is very often at 100 %, which 
> basically makes the computer pretty much useless for anything (else).
> The server's specs are: XP SP3, Celeron 2.4 GHz, 1 GB RAM. Too weak for 0.7x?
Firstly, if your computer is unstable when running at 100% CPU for long 
periods, your computer is broken and should be RMA'ed. :)

Seriously, the above specs should be adequate, depending on what else is 
running. What else is running? Freenet should not, generally speaking, run 100% 
CPU for extended periods. However, LIKE ALMOST ANY OTHER APPLICATION, 100% CPU 
for brief periods is inevitable. Windows' scheduling priority code, combined 
with the fact that Freenet runs at BELOW_NORMAL priority class, should ensure 
that this does not interfere in any significant way with other applications; 
unless you are running things other than freenet at low priorities, it is more 
likely that a general slowdown is caused by the constant and fairly heavy disk 
I/O freenet does.

However, if Freenet is doing a lot (e.g. lots of downloads queued) and has too 
low a memory limit, that can push it into constant 100% CPU, constantly seeking 
out the last few bytes. That's probably the most common problem. You might like 
to increase the max memory limit (in wrapper.conf). However, this means less 
memory for your other apps. Note that Freenet uses a bit of memory on top of 
the memory limit - some for threads (you can limit the number of threads at 
some performance cost), some for the JVM itself, and some for the bloom filters 
for the datastore (1MB per 2GB).

Attachment: signature.asc
Description: This is a digitally signed message part.

Support mailing list
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Reply via email to