loss of cause??

[STIRLING, SCOTT]

I thought there was a JSR to standardize on the log format, or had heard that a couple years ago at JavaOne in SFO. Might've been at one of Kirk's presentations, if I recall! I remember asking about it. IBM, for example, at least has an XML format to the JVM GC log so you can write tools around a parseable DOM -- but Sun, for years and years, has had this ad hoc, increasingly difficult to parse, loosely structured log format that changes regularly. It's kind of out of control ...

FWIW,
Scott Stirling
AT&T, Boston

-----Original Message-----
From: hotspot-gc-dev-bounces at openjdk.java.net [mailto:hotspot-gc-dev-bounces at openjdk.java.net] On Behalf Of Kirk Pepperdine
Sent: Thursday, October 10, 2013 4:20 PM
To: Mikael Gerdin
Cc: hotspot-gc-dev at openjdk.java.net openjdk.java.net
Subject: Re: loss of cause??

Hi Mikael,

Thanks for the response. I'm aware of the flag. But more importantly I want, and I think I can speak for other, is for the log formats to remain stable. These things are painful enough to parse without having to compensate for variations not only across builds but across different OS'es.

Kind regards,
Kirk Pepperdine

On 2013-10-10, at 4:00 PM, Mikael Gerdin <mikael.gerdin at oracle.com> wrote:

> On 10/10/2013 03:46 PM, Pierre Laporte wrote:
>> This is also true for Linux :
>> http://www.pingtimeout.fr/2013/08/gc-log-format-change.html
> 
> If you want the GC causes for all gcs in 7u40 please use -XX:+PrintGCCause
> 
> /Mikael
> 
>> 
>> 
>> On Thu, Oct 10, 2013 at 3:30 PM, Kirk Pepperdine <kirk at kodewerk.com
>> <mailto:kirk at kodewerk.com>> wrote:
>> 
>>    Hi all,
>> 
>>    I've just noticed this in a GC log.
>> 
>>    642.015: [Full GC [PSYoungGen: 1904K->0K(79360K)] [ParOldGen:
>>    337438K->338524K(442368K)] 339342K->338524K(521728K) [PSPermGen:
>>    13239K->13239K(21504K)], 4.4455297 secs] [Times: user=7.94 sys=0.02,
>>    real=4.44 secs]
>> 
>> 
>>    This collection was triggered by a call to System.gc().
>> 
>>    More info
>>    D:\dev\>"C:\Program Files\Java\jdk1.7.0_40\bin\java" -version
>>    java version "1.7.0_40"
>>    Java(TM) SE Runtime Environment (build 1.7.0_40-b43)
>>    Java HotSpot(TM) 64-Bit Server VM (build 24.0-b56, mixed mode)
>> 
>>    Flags used -Xloggc:gc.log -XX:+PrintGCDetails
>> 
>> 
>>    Same app running in OSX produces;
>> 
>>    2.435: [Full GC (System) [PSYoungGen: 32K->0K(24064K)] [ParOldGen:
>>    3666K->3666K(54976K)] 3698K->3666K(79040K) [PSPermGen:
>>    9620K->9620K(22272K)], 0.0311970 secs] [Times: user=0.06 sys=0.00,
>>    real=0.03 secs]
>> 
>>    Obviously triggered by a full GC
>> 
>>    flash-2:kirk$ java -version
>>    java version "1.7.0_40"
>>    Java(TM) SE Runtime Environment (build 1.7.0_40-b43)
>>    Java HotSpot(TM) 64-Bit Server VM (build 24.0-b56, mixed mode)
>> 
>>    Flags used -Xloggc:gc.log -XX:+PrintGCDetails
>> 
>>    I've not checked for this on Linux yet.
>> 
>>    So, it looks as though the (System) has been lost on windows?
>> 
>>    Regards,
>>    Kirk
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> --
>> 
>> Pierre Laporte
>> @pingtimeout <https://twitter.com/pingtimeout>
>> pierre at pingtimeout.fr <mailto:pierre at pingtimeout.fr>
>> http://www.pingtimeout.fr/
>