other nodes gc log looks like this: [GC 99623.647: [ParNew: 284027K->6782K(314560K), 0.0136660 secs] 818985K->541887K(1013632K), 0.0137850 secs] [Times: user=0.10 sys=0.00, real=0.02 secs] [GC 99636.522: [ParNew: 286398K->9306K(314560K), 0.0149200 secs] 821503K->544533K(1013632K), 0.0150510 secs] [Times: user=0.11 sys=0.00, real=0.01 secs]
I think these are minor GC, CMS mark-sweep is not minor GC On Sun, Mar 13, 2016 at 7:57 PM, Manikumar Reddy <manikumar.re...@gmail.com> wrote: > Hi, > > These logs are minor GC logs and they look normal. Look for the word 'Full' > for full gc log details. > > > > > On Sun, Mar 13, 2016 at 3:06 PM, li jinyu <disorder...@gmail.com> wrote: > > > I'm using Kafka 0.8.1.1, have 10 nodes in a cluster, all are started with > > default command: > > ./bin/kafka-server-start.sh conf/server.properties > > > > but yesterday, I found that two nodes kept full GC(every 1-3 seconds), > > although there was still enough memory: > > > > 99546.435: [GC [1 CMS-initial-mark: 532714K(699072K)] 538916K(1013632K), > > 0.0099270 secs] [Times: user=0.01 sys=0.00, real=0.01 secs] > > 99546.446: [CMS-concurrent-mark-start] > > 99546.520: [CMS-concurrent-mark: 0.074/0.074 secs] [Times: user=0.21 > > sys=0.03, real=0.08 secs] > > 99546.520: [CMS-concurrent-preclean-start] > > 99546.525: [CMS-concurrent-preclean: 0.005/0.005 secs] [Times: user=0.01 > > sys=0.00, real=0.00 secs] > > 99546.525: [CMS-concurrent-abortable-preclean-start] > > 99547.348: [CMS-concurrent-abortable-preclean: 0.822/0.823 secs] [Times: > > user=1.53 sys=0.31, real=0.83 secs] > > 99547.350: [GC[YG occupancy: 158101 K (314560 > > K)]2016-03-12T20:19:58.597+0800: 99547.350: [GC 99547.350: [ParNew: > > 158101K->5170K(314560K), 0.0189700 secs] 690816K->538498K(1013632K), > > 0.0190720 secs] [Times: user=0.11 sys=0.00, real=0.02 secs] > > 99547.369: [Rescan (parallel) , 0.0099240 secs]99547.379: [weak refs > > processing, 0.0000090 secs]99547.379: [class unloading, 0.0028860 > > secs]99547.382: [scrub symbol table, 0.0015400 secs]99547.383: [scrub > > string table, 0.0001640 secs] [1 CMS-remark: 533327K(699072K)] > > 538498K(1013632K), 0.0353890 secs] [Times: user=0.18 sys=0.00, real=0.03 > > secs] > > 99547.386: [CMS-concurrent-sweep-start] > > 99547.421: [CMS-concurrent-sweep: 0.035/0.035 secs] [Times: user=0.08 > > sys=0.01, real=0.04 secs] > > 99547.421: [CMS-concurrent-reset-start] > > 99547.426: [CMS-concurrent-reset: 0.005/0.005 secs] [Times: user=0.01 > > sys=0.01, real=0.00 secs] > > 99549.132: [GC 99549.132: [ParNew: 284786K->4184K(314560K), 0.0158450 > secs] > > 816880K->536891K(1013632K), 0.0159570 secs] [Times: user=0.08 sys=0.00, > > real=0.01 secs] > > > > > > is this caused by small head(1G by default), or consumer lag? > > > > I tried to use jmap to check the heap, but failed to attach to the java > > process. > > how to find the root cause? > > > > -- > > Don't schedule every day, make them disorder. > > > -- Don't schedule every day, make them disorder.
