lei.xu created TS-2847: -------------------------- Summary: cpu usage like a roller coaster occasionally Key: TS-2847 URL: https://issues.apache.org/jira/browse/TS-2847 Project: Traffic Server Issue Type: Test Components: Network Reporter: lei.xu
The data of tsar is shown below: Time ---cpu-- ---mem-- ---tcp-- -----traffic---- 27/05/14-08:05 6.74 70.56 5.48 1.2M 5.4M 27/05/14-08:10 5.35 70.63 5.41 890.9K 4.4M 27/05/14-08:15 4.57 70.76 5.88 990.8K 4.9M 27/05/14-08:20 4.51 70.88 7.10 833.0K 5.2M 27/05/14-08:25 9.93 71.08 6.59 779.6K 4.1M 27/05/14-08:30 6.60 71.12 6.02 865.8K 4.8M 27/05/14-08:35 42.02 71.52 12.89 582.1K 1.9M 27/05/14-08:40 43.41 19.30 13.48 989.3K 1.4M 27/05/14-08:45 4.66 21.08 6.61 1.0M 4.9M 27/05/14-08:50 4.97 21.72 6.12 1.0M 5.7M 27/05/14-08:55 4.67 22.27 5.58 1.0M 5.3M 27/05/14-09:00 4.77 22.90 7.58 938.2K 4.9M 27/05/14-09:05 4.82 23.17 7.04 1.0M 5.3M 27/05/14-09:10 4.43 23.62 5.84 977.4K 4.8M 27/05/14-09:15 4.23 24.10 6.28 948.8K 4.8M 27/05/14-09:20 4.09 24.59 5.23 966.8K 4.9M 27/05/14-09:25 3.96 25.02 5.98 767.5K 4.3M 27/05/14-09:30 4.07 25.44 6.79 768.7K 4.8M 27/05/14-09:35 4.18 25.87 6.40 904.9K 4.9M May 27 08:38:00 traffic_cop[28952]: server heartbeat failed [2] when the cpu usage is increasing suddenly, gdb the process of traffic_server. 182 pthread_cond_timedwait@@GLIBC_2.3.2 27 epoll_wait 8 nanosleep 4 accept 2 pread64 1 msync when the cpu usage is normal, gdb the process of traffic_server. 91 pthread_cond_timedwait@@GLIBC_2.3.2 14 epoll_wait 4 nanosleep 2 accept 1 pread64 My English is poor, sorry. -- This message was sent by Atlassian JIRA (v6.2#6252)