Add linebreak for CMS abort?

Bernd Eckenfels bernd.eckenfels at googlemail.com
Fri Oct 12 05:46:50 UTC 2012


Hello,

what do you think, does it make sense to add a linebreak at the following
aborted message for verboseGC? (timestamps truncated, system times
removed).

The " CMS: abort preclean due to time " has no timestamp and especially no
linebreak at the end.

09:57:38.278: [GC [ParNew: 268442K->6031K(2516544K), 0.0096560 secs]
268442K->6031K(8108992K), 0.0097540 secs]
09:57:38.288: [GC [1 CMS-initial-mark: 0K(5592448K)] 50770K(8108992K),
0.0186090 secs]
09:57:38.349: [CMS-concurrent-mark: 0.011/0.041 secs]
09:57:38.380: [CMS-concurrent-preclean: 0.031/0.031 secs]
    CMS: abort preclean due to time 09:57:48.451:
[CMS-concurrent-abortable-preclean: 9.189/10.071 secs]
09:57:48.452: [GC[YG occupancy: 789050 K (2516544
K)]2012-10-11T09:57:48.452-0800: [GC [ParNew: 789050K->48388K(2516544K),
0.0730430 secs] 789050K->54825K(8108992K), 0.0731570 secs]
[Rescan (parallel) , 0.0166130 secs][weak refs processing, 0.0000180
secs][class unloading, 0.0147850 secs][scrub symbol & string tables,
0.0083180 secs] [1 CMS-remark: 6437K(5592448K)] 54825K(8108992K),
0.1143460 secs]
09:57:48.582: [CMS-concurrent-sweep: 0.009/0.015 secs]
09:57:48.643: [CMS-concurrent-reset: 0.060/0.062 secs]
09:58:08.776: [GC [ParNew: 2285316K->132294K(2516544K), 0.1732660 secs]
2291753K->188592K(8108992K), 0.1734070 secs]

A second option to making this additional line fit into the line schema
would be to actually add some statistics inside the record:

old:
    CMS: abort preclean due to time 09:57:48.451:
[CMS-concurrent-abortable-preclean: 9.189/10.071 secs]

new1:
09:57:38.380: [CMS-concurrent-preclean: 0.031/0.031 secs]
(09:57:48.451:) CMS: abort preclean due to time
09:57:48.451: [CMS-concurrent-abortable-preclean: 9.189/10.071 secs]

new2:
09:57:38.380: [CMS-concurrent-preclean: 0.031/0.031 secs]
09:57:48.451: [CMS-concurrent-abortable-preclean-aborted: 9.189/10.071
secs]

Greetings
Bernd

-- 
https://plus.google.com/u/1/108084227682171831683/about



More information about the hotspot-gc-dev mailing list