JEP 271: Unified GC Logging - Third pre-review
bengt.rutisson at oracle.com
Tue Nov 17 07:51:59 UTC 2015
Hi Kirk and Bernd,
I talked to Marcus Larsson about this. He filed this bug to track the work:
Add headers to log outputs in unified logging
I can't promise anything, but hopefully we'll get these headers soon.
On 2015-11-15 09:31, Kirk Pepperdine wrote:
> The header that has been showing up, for example...
> Java HotSpot(TM) 64-Bit Server VM (25.60-b23) for linux-amd64 JRE (1.8.0_60-b27), built on Aug 4 2015 12:19:40 by "java_re" with gcc 4.3.0 20080428 (Red Hat 4.3.0-8)
> Memory: 4k page, physical 125650660k(120686628k free), swap 0k(0k free)
> CommandLine flags: -XX:+AlwaysPreTouch -XX:ConcGCThreads=8 -XX:InitialHeapSize=53687091200 -XX:InitiatingHeapOccupancyPercent=35 -XX:MaxGCPauseMillis=500 -XX:MaxHeapSize=53687091200 -XX:MaxMetaspaceSize=536870912 -XX:MetaspaceSize=536870912 -XX:ParallelGCThreads=15 -XX:+ParallelRefProcEnabled -XX:+PrintAdaptiveSizePolicy -XX:+PrintGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+UseG1GC
> Is the header that I’m completely missing in this log file. I can currently derive a lot of this information from patterns the the GC logs such as JVM version and other bits here and there but I can’t get to all of it. Having it in the GC log IME, is hugely useful, having it in the headers of rotated logs would also be hugely useful but would be something new.
>> On Nov 13, 2015, at 2:13 PM, Bengt Rutisson <bengt.rutisson at oracle.com> wrote:
>> Hi Bernd,
>> ----- ecki at zusammenkunft.net wrote:
>>> Hello Bengt,
>>> Looks good, one question on the "first line", you write out the gc
>>> policy in use, will this be repeatet every time the file is rolled
>>> In the current GC logs the VM version and option details had been
>>> added. I think this was imensly usefull to have all the important bits
>>> at the beginning of each logfile. Can you include that as part of gc
>>> tags? (Datestamp, pid, gc selected, vm uptime, log generation, command
>>> line arguments and maybe even OS memory details)
>> I asked Marcus Larsson about this. There is currently no support in the unified logging framework for writing a header when the files are rotated. We realize that this was very useful and it is something we should probably add.
>> So, the answer to your question is no at the moment, but we'll try to add support for this.
>>> -----Original Message-----
>>> From: Bengt Rutisson <bengt.rutisson at oracle.com>
>>> To: "hotspot-gc-dev at openjdk.java.net"
>>> <hotspot-gc-dev at openjdk.java.net>
>>> Sent: Fr., 13 Nov. 2015 10:19
>>> Subject: JEP 271: Unified GC Logging - Third pre-review
>>> Hi everyone,
>>> Here is an updated pre-review for the work for JEP 271 Unified GC
>>> Now it includes all the logging that I plan on converting for the
>>> Here are examples from the latest prototype:
>>> The webrev for the changes is available here:
>>> Some changes that have been made:
>>> - GC id is now logged as "GC(5)" instead of the previous "GC#5"
>>> - The CMS and G1 concurrent logging have been homogenized to look
>>> similar. All concurrent phases are start with "Concurrent" and now
>>> include timestamps.
>>> - The paused phases now all start with "Pause".
>>> - The log messages from the different collectors now look more
>>> For example "Pause Young" is used for young GCs for all collectors
>>> instead of for example "Young pause" for G1 and "Pause. ParNew" for
>>> - I've added a line at the start of the log to clearly state what GC
More information about the hotspot-gc-dev