Fwd: Question jruby and java heap memory message

Ruby JRuby
21

As far as I know, it is not possible to change max heap size after you
have started the JVM. However, you can control how much heap JVM
initially allocates when it starts (it does not have to go for the max
as soon as it starts) through -J-Xms####m so that you can give a bigger
number for max as a cap, but a smaller number to start with so that if
it does not need that much, it won’t grep the max from the operating
system.

Cheers,
Chiaming H.

----- Original Message ----
From: Roger P. [email protected]
To: [email protected]
Sent: Thu, February 25, 2010 9:56:45 AM
Subject: [jruby-user] Re: Fwd: Question jruby and java heap memory
message

Try increasing the max heap size as suggested:

Error: Your application used more memory than the safety cap of 500m.
Specify -J-Xmx####m to increase it (#### = cap size in MB).
Specify -w for full OutOfMemoryError stack trace

I assume it’s impossible to change these specifications
“programmatically” after you’ve started java?
-r

Posted via http://www.ruby-forum.com/.

To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email

To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email
22

yep, also I’m using jmeter as second option, but I want to run that with
other tool like > selenium rc, watir, jruby or celerity :S

your comment makes me wonder: what’s your final goal ? Is it to achieve
some
screen-scraping ? Is it to simulate a load to do performance testing on
a
website ? … ?

– Thibaut

23

mmmmm job queue, but that won’t work if I want 50 concurrent users
connected at same time

If you want to ensure the system remains stable instead of exploding
once
one more user is there, queuing allows you to keep a (more) stable
throughput.

See the following graph and article to ilustrate my point:

This allows the system to run at the pace it is able to run. You can
work on
optimizing the individual job performance as well, of course.

This also allows to scale out (ie: use another process or another
machine to
handle more jobs concurrently). I’m currently using Resque
GitHub - defunkt/resque: Moved to resque/resque for that purpose.

hope this helps :slight_smile:

– Thibaut

24

ok I will check it.
thanks,
moises

On Fri, Feb 26, 2010 at 2:21 AM, Thibaut Barrère