users@glassfish.java.net

Re: Glassfish 3.1 is much slower to deploy

From: Andreas Loew <Andreas.Loew_at_oracle.com>
Date: Tue, 12 Apr 2011 16:55:20 +0200

Hi Kumar,

even if it is not an regression in 3.1 over 3.0 but only over 2.1.1
(i.e. the issue was already introduced with 3.0), I think this
definitely is worth creating a GF issue (or a formal Service Request in
case the original poster has a license/support contract).

The fact that deployment time increases by a factor of 9 when just
adding @RolesAllowed annotations clearly seems to point to a
performance-related code flaw...

Best regards,

Andreas


Am 12/04/11 05:37, schrieb Kumar Jayanti:
> On 12-Apr-2011, at 8:19 AM,forums_at_java.net wrote:
>
>> > Hi, I made a test case with 800 methods called hello_000 to hello_800 and it
>> > deployed quickly. each looked like public String hello_000() { return
>> > "hello"; } I then added @RolesAllowed on about a third of the methods and
>> > this is what caused the slowdown when when 20 000 millseconds to 180 000
>> > milliseconds eg at top of class @DeclareRoles( {"Role1", "Role2", "Role3",
>> > "Role4", "Role5", "Role6"}) @RolesAllowed( {"Role1", "Role2", "Role3",
>> > "Role4", "Role5", "Role6"}) @Stateless public class TestStateless implements
>> > TestStatelessRemote { Now some methods look like @RolesAllowed( {"Role1",
>> > "Role2" }) public String hello_001() { return "hello"; } My development
>> > system is not connected to the internet so sending you the test case is
>> > difficult But it seems that the roles processing is the slow part
> If you see this as a regression over V3.0 then we need to worry about it. Can you confirm ...
>
> regards,
> kumar

-- 
Andreas Loew | Senior Java Architect
Oracle Advanced Customer Services
ORACLE Deutschland B.V. & Co. KG