I see only one file opened on XP:
java.exe pid: 5384 C:\temp\test.jar
kedar wrote:
> This is intriguing. I somehow feel that it is the lsof output on
> non-Solaris
> platforms that shows the same file listed twice. For example, Bill, in
> your case
> on Mac, the file descriptor (FD) is reported as 304r (fd/read) and txt
> (program code and data) for a file. For me, on Linux, for a similar
> server
> run, lsof shows two rows for a file with FD being mem (memory mapped
> file) and 304r (fd/read).
> (It does appear however that each entry of lsof output consumes an FD).
>
> So could this really be a JDK bug?
>
>
> Bill Shannon wrote:
>
>> Interesting. I tried it and I see only one open jar file on Solaris 10.
>> I wonder what it is about the other platforms that's causing it to open
>> the file more than once?
>>
>> Are you going to file a JDK bug? If so, let us know.
>>
>>
>> Richard S. Hall wrote on 10/13/09 3:44 AM:
>>
>>> Well, here is the test program I wrote:
>>>
>>> package test;
>>> import java.util.jar.JarFile;
>>> public class Main {
>>> public static void main(String[] args) throws Exception {
>>> JarFile jar = new JarFile("test.jar");
>>> System.out.println("+++ AFTER");
>>> Thread.sleep(120000);
>>> }
>>> }
>>>
>>> Under Windows test.jar appears to be opened twice using handle.exe:
>>>
>>> java.exe pid: 3992 HEAVYWEIGHT-XP\rickhall
>>> 694: File (RW-) C:\Documents and
>>> Settings\rickhall\Projects\tmp\test.jar
>>> 698: File (RW-) C:\Documents and
>>> Settings\rickhall\Projects\tmp\test.jar
>>>
>>> Under Snow Leopard test.jar appears to be opened twice using lsof:
>>>
>>> java 27559 rickhall 30r REG 14,2 938 35110250
>>> /Users/rickhall/tmp/test.jar
>>> java 27559 rickhall 31r REG 14,2 938 35110250
>>> /Users/rickhall/tmp/test.jar
>>>
>>> Under Linux test.jar appears to be opened twice using lsof:
>>>
>>> java 2527 rickhall 4r REG 253,0 949 508106
>>> /home/rickhall/test.jar
>>> java 2527 rickhall 5r REG 253,0 949 508106
>>> /home/rickhall/test.jar
>>>
>>> So, it doesn't appear to be specific to the Mac. Part of me wonders if
>>> this is just some confusion in how these utilities report open file...
>>>
>>> I actually tried to test it on OpenSolaris too. That was a little more
>>> complicated and I don't know if I was successful or not, but in that
>>> case it seems like I only saw one open file.
>>>
>>> -> richard
>>>
>>> On 10/13/09 0:19, Richard S. Hall wrote:
>>>
>>>> Ok, before reporting this, I will check on a couple other platforms
>>>> (e.g., Linux and Windows) to make sure it isn't Mac OS specific.
>>>>
>>>> -> richard
>>>>
>>>> On 10/12/09 23:42, Jeanfrancois Arcand wrote:
>>>>
>>>>>
>>>>> Bill Shannon wrote:
>>>>>
>>>>>> Richard S. Hall wrote on 10/11/09 4:41 AM:
>>>>>>
>>>>>>> I set a break point where Felix opens the the first bundle JAR file
>>>>>>> using JarFile. Before creating the JarFile there was no open bundle
>>>>>>> JAR
>>>>>>> files, after it there were two. I even created a simple test
>>>>>>> program
>>>>>>> that only opens a JarFile in its main method and sleeps for 30
>>>>>>> seconds.
>>>>>>> This is what I saw:
>>>>>>>
>>>>>>> java 14129 rickhall 30r REG 14,2 846
>>>>>>> 43270278
>>>>>>> /Users/rickhall/tmp/jar/jar.jar
>>>>>>> java 14129 rickhall 31r REG 14,2 846
>>>>>>> 43270278
>>>>>>> /Users/rickhall/tmp/jar/jar.jar
>>>>>>>
>>>>>>> So, it seems JarFile always reports two open files. Not sure why.
>>>>>>
>>>>>> So, I guess it's a bug, or feature, of the JDK.
>>>>>>
>>>>>> Maybe you should file a bug and see what they say? You might want
>>>>>> to search the bug database first.
>>>>>
>>>>> FYI I've talked with the SE team and it seems it is not a known issue
>>>>> (at least based on this email thread). In any case we should document
>>>>> the fact that GF needs more than the usual default fd to runs
>>>>> properly, because we will ends up with tons of post on 'GlassFish
>>>>> leaks fd' :-)
>>>>>
>>>>> -- Jeanfrancois
>>>>>
>>>>>
>>>>>> ---------------------------------------------------------------------
>>>>>>
>>>>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>>>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>>>>
>>>>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>>>
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>
--
Byron Nevins Work 408-276-4089, Home 650-359-1290, Cell 650-784-4123 - Sun Microsystems, Inc.