Java - Azureus memory usage

[Garry Harthill]

top - 14:24:28 up 5 days,  3:41,  5 users,  load average: 3.78, 3.61, 3.27
Tasks: 117 total,   1 running, 116 sleeping,   0 stopped,   0 zombie
Cpu(s): 57.0% us, 10.3% sy,  0.0% ni, 31.1% id,  1.3% wa,  0.3% hi,  0.0% si
Mem:   1027288k total,  1002144k used,    25144k free,    39564k buffers
Swap:  1020116k total,      940k used,  1019176k free,   405756k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
13132 garry     15   0  966m 213m  37m S 50.6 21.3 227:27.00 gij

213mb memory and 50% CPU seems a bit excessive.

azureus-2.4.0.0-0.20060209cvs_1.fc5
glib-java-0.2.3-1.2
cairo-java-1.0.2-0.2
libgconf-java-2.12.1-2.2
libgtk-java-2.8.3-1.2
java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh

Anyone else seeing this?

Garry

[Andrew Haley]

Garry Harthill writes:
>  top - 14:24:28 up 5 days,  3:41,  5 users,  load average: 3.78, 3.61,
> 3.27
>  Tasks: 117 total,   1 running, 116 sleeping,   0 stopped,   0 zombie
>  Cpu(s): 57.0% us, 10.3% sy,  0.0% ni, 31.1% id,  1.3% wa,  0.3% hi,  0.0% si
>  Mem:   1027288k total,  1002144k used,    25144k free,    39564k buffers
>  Swap:  1020116k total,      940k used,  1019176k free,   405756k cached
>  
>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>  13132 garry     15   0  966m 213m  37m S 50.6 21.3 227:27.00 gij
>  
>  213mb memory and 50% CPU seems a bit excessive.
>  
>  azureus-2.4.0.0-0.20060209cvs_1.fc5
>  glib-java-0.2.3-1.2
>  cairo-java-1.0.2-0.2
>  libgconf-java-2.12.1-2.2
>  libgtk-java-2.8.3-1.2
>  java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh
>  
>  Anyone else seeing this? 
When you start azureus, use 'gij -verbose:class'.  See if any of the
classes are loaded 'bytecode'.

Andrew.

[Ralf Ertzinger]

Hi

On Tue, 28 Feb 2006 14:41:57 +0000, Andrew Haley wrote:

>   > Anyone else seeing this?
>  
>  When you start azureus, use 'gij -verbose:class'.  See if any of the
>  classes are loaded 'bytecode'. 
2929 unique classes. I'm impressed. Some of these are bytecode, indeed.
Complete list is attached.

[Anthony Green]

On Tue, 2006-02-28 at 16:22 +0100, Ralf Ertzinger wrote:
>  Hi
>  
>  On Tue, 28 Feb 2006 14:41:57 +0000, Andrew Haley wrote:
>  
>  >  > Anyone else seeing this?
>  > 
>  > When you start azureus, use 'gij -verbose:class'.  See if any of the
>  > classes are loaded 'bytecode'.
>  
>  2929 unique classes. I'm impressed. Some of these are bytecode, indeed.
>  Complete list is attached. 
Thanks.  I see we're using a slow interpreter for the crypto code.  This
is a bad idea.  I've filed two bugs...

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=183451
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=183452

AG

[Andrew Haley]

Andrew Haley writes:
>  Garry Harthill writes:
>   > top - 14:24:28 up 5 days,  3:41,  5 users,  load average: 3.78, 3.61, 3.27
>   > Tasks: 117 total,   1 running, 116 sleeping,   0 stopped,   0 zombie
>   > Cpu(s): 57.0% us, 10.3% sy,  0.0% ni, 31.1% id,  1.3% wa,  0.3% hi,  0.0% si
>   > Mem:   1027288k total,  1002144k used,    25144k free,    39564k buffers
>   > Swap:  1020116k total,      940k used,  1019176k free,   405756k cached
>   > 
>   >   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>   > 13132 garry     15   0  966m 213m  37m S 50.6 21.3 227:27.00 gij
>   > 
>   > 213mb memory and 50% CPU seems a bit excessive.
>   > 
>   > azureus-2.4.0.0-0.20060209cvs_1.fc5
>   > glib-java-0.2.3-1.2
>   > cairo-java-1.0.2-0.2
>   > libgconf-java-2.12.1-2.2
>   > libgtk-java-2.8.3-1.2
>   > java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh
>   > 
>   > Anyone else seeing this?
>  
>  When you start azureus, use 'gij -verbose:class'.  See if any of the
>  classes are loaded 'bytecode'. 
It looks like this guess was wrong.  The fact that others are seeing
degraded functionality as well as degraded performancew suggests that
something major is wrong.

Andrew.

[Dominik 'Rathann' Mierzejewski]

On Tuesday, 28 February 2006 at 15:30, Garry Harthill wrote:
>  top - 14:24:28 up 5 days,  3:41,  5 users,  load average: 3.78, 3.61,
> 3.27
>  Tasks: 117 total,   1 running, 116 sleeping,   0 stopped,   0 zombie
>  Cpu(s): 57.0% us, 10.3% sy,  0.0% ni, 31.1% id,  1.3% wa,  0.3% hi,  0.0% si
>  Mem:   1027288k total,  1002144k used,    25144k free,    39564k buffers
>  Swap:  1020116k total,      940k used,  1019176k free,   405756k cached
>  
>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>  13132 garry     15   0  966m 213m  37m S 50.6 21.3 227:27.00 gij
>  
>  213mb memory and 50% CPU seems a bit excessive.
>  
>  azureus-2.4.0.0-0.20060209cvs_1.fc5
>  glib-java-0.2.3-1.2
>  cairo-java-1.0.2-0.2
>  libgconf-java-2.12.1-2.2
>  libgtk-java-2.8.3-1.2
>  java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh
>  
>  Anyone else seeing this? 
What I'm seeing is deficient functionality. FE version seems to be able
to handle only one download at a time and at far lower speed, i.e.
only one of all active downloads actually shows any "Down speed",
all others are at zero. It is also much slower in connecting to tracker
and finding out how many peers and seeds there are.
The binary from azureus.sf.net coupled with jpackage.org's sun java rpm
works flawlessly and handles as many concurrent downloads as I want.

Regards,
R.

-- 
       RPM repository for Fedora Core http://rpm.greysector.net/
mpg321, xmp, faad2, lame, mad, *mplayer*, rdesktop, tin, xvid, mks, mutt
"Faith manages."
        -- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"

[Anthony Green]

On Tue, 2006-02-28 at 18:43 +0100, Dominik 'Rathann' Mierzejewski wrote:
>  
>  What I'm seeing is deficient functionality. FE version seems to be
>  able
>  to handle only one download at a time and at far lower speed, i.e.
>  only one of all active downloads actually shows any "Down speed",
>  all others are at zero. 
I'm not seeing this, but please file a bug report and we can try to
solve it.

Thanks,

AG

[Anthony Green]

On Tue, 2006-02-28 at 14:30 +0000, Garry Harthill wrote:
>  213mb memory and 50% CPU seems a bit excessive.
>  
>  azureus-2.4.0.0-0.20060209cvs_1.fc5
>  glib-java-0.2.3-1.2
>  cairo-java-1.0.2-0.2
>  libgconf-java-2.12.1-2.2
>  libgtk-java-2.8.3-1.2
>  java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh
>  
>  Anyone else seeing this? 
It's big for me, but not using so much CPU.  Please file a bug report @
http://bugzilla.redhat.com.

Thanks,

AG

[Garry Harthill]

On 01/03/06, Anthony Green <green(a)redhat.com&gt; wrote:
>  On Tue, 2006-02-28 at 14:30 +0000, Garry Harthill wrote:
>  > 213mb memory and 50% CPU seems a bit excessive.
>  >
>  > azureus-2.4.0.0-0.20060209cvs_1.fc5
>  > glib-java-0.2.3-1.2
>  > cairo-java-1.0.2-0.2
>  > libgconf-java-2.12.1-2.2
>  > libgtk-java-2.8.3-1.2
>  > java-1.4.2-gcj-compat-1.4.2.0-40jpp_80rh
>  >
>  > Anyone else seeing this?
>
>  It's big for me, but not using so much CPU.  Please file a bug report @
>  http://bugzilla.redhat.com.
>  
CPU usage varies. This was maybe a particularly high example. Anyway, bugged:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=183461

[Pete Zaitcev]

On Tue, 28 Feb 2006 14:30:35 +0000, "Garry Harthill" <gazzerh(a)gmail.com&gt;
wrote:

>  213mb memory and 50% CPU seems a bit excessive. 
>  Anyone else seeing this? 
Nope, mine is at about 53MB with Fedora torrents. It eats a bit of CPU,
but I run a VIA C3. Heck, ssh eats CPU on that wimpy chip.

-- Pete

[Andrew Haley]

Pete Zaitcev writes:
>  On Tue, 28 Feb 2006 14:30:35 +0000, "Garry Harthill"
> <gazzerh(a)gmail.com&gt; wrote:
>  
>  > 213mb memory and 50% CPU seems a bit excessive.
>  
>  > Anyone else seeing this?
>  
>  Nope, mine is at about 53MB with Fedora torrents. It eats a bit of CPU,
>  but I run a VIA C3. Heck, ssh eats CPU on that wimpy chip. 
OH, right.  Thanks for that.  I wonder what causes the difference.

Andrew.