RFR: 7164841: Improvements to the GC log file rotation

Yumin Qi yumin.qi at oracle.com
Sat Aug 31 03:09:45 UTC 2013

Sample of the file head after new patch:

2013-08-30 14:35:25 GC log file created 
Java HotSpot(TM) Server VM (25.0-b46-internal-fastdebug) for linux-x86 
JRE (1.8.0), built on Aug 30 2013 14:30:10 by "yqi" with gcc 4.4.6 
20120305 (Red Hat 4.4.6-4)
Memory: 4k page, physical 63293852k(39066172k free), swap 
32104444k(32104444k free)
CommandLine flags: -XX:GCLogFileSize=16384 -XX:InitialHeapSize=67108864 
-XX:MaxHeapSize=1073741824 -XX:MaxNewSize=357892096 
-XX:MaxTenuringThreshold=6 -XX:NumberOfGCLogFiles=3 -XX:OldPLABSize=16 
-XX:+PrintGC -XX:+PrintGCDetails -XX:+PrintGCTimeStamps 
-XX:+ShowMessageBoxOnError -XX:+UseConcMarkSweepGC 
-XX:+UseGCLogFileRotation -XX:+UseParNewGC
120.274: [GC (Allocation Failure) 120.274: [ParNew: 17305K->2K(19648K), 
0.0045280 secs] 179491K->162188K(289960K), 0.0047320 secs] [Times: 
user=0.00 sys=0.00, real=0.01 secs]
120.282: [GC (Allocation Failure) 120.282: [ParNew: 17305K->2K(19648K), 
0.0043440 secs] 179491K->162188K(289960K), 0.0046470 secs] [Times: 
user=0.00 sys=0.00, real=0.00 secs]
120.290: [GC (Allocation Failure) 120.290: [ParNew: 17305K->2K(19648K), 
0.0042520 secs] 179491K->162188K(289960K), 0.0045540 secs] [Times: 
user=0.00 sys=0.00, real=0.00 secs]
120.298: [GC (Allocation Failure) 120.298: [ParNew: 17305K->2K(19648K), 
0.0043110 secs] 179491K->162188K(289960K), 0.0046080 secs] [Times: 
user=0.01 sys=0.00, real=0.00 secs]
120.306: [GC (Allocation Failure) 120.306: [ParNew: 17305K->2K(19648K), 
0.0044750 secs] 179491K->162188K(289960K), 0.0047500 secs] [Times: 
user=0.01 sys=0.00, real=0.01 secs]
120.314: [GC (Allocation Failure) 120.314: [ParNew: 17305K->2K(19648K), 
0.0046360 secs] 179491K->162188K(289960K), 0.0048120 secs] [Times: 
user=0.01 sys=0.00, real=0.01 secs]


On 8/29/2013 4:38 PM, Bernd Eckenfels wrote:
> Am 29.08.2013, 23:37 Uhr, schrieb Yumin Qi <yumin.qi at oracle.com>:
>>    Thanks for your review. I will send out a new webrev soon.
>>    For your concern, see embedded answers.
> Could you maybe also sent a sample header snippet?
> I raised concern that each logfile segment should have enough 
> information in the files to allow easy support (vm version, os 
> version, ram/swap size, "PrintCommandLineFlags"(at least heap sizing 
> and gc type), pid, vm-start-time, logfile-start-time)-  But I think 
> the mail did not get trough the list.
> It would be good if each rotated logfile contains enough information 
> to find the absolute time of each event and answer all questions 
> supporters will have to identify the actual run. This will also help 
> the work of this list looking at erratic gc logs!
> Greetings
> Bernd

More information about the hotspot-gc-dev mailing list