process.sh 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125
  1. #! /bin/bash
  2. env=$1 # 环境
  3. if [ ${env} = "dev" ];then
  4. piaoquan_crawler_dir=/Users/wangkun/Desktop/crawler/piaoquan_crawler/
  5. profile_path=/etc/profile
  6. python=python3
  7. log_path=${piaoquan_crawler_dir}main/main_logs/process-$(date +%Y-%m-%d).log
  8. else
  9. piaoquan_crawler_dir=/data5/piaoquan_crawler/
  10. profile_path=/etc/profile
  11. python=python
  12. log_path=${piaoquan_crawler_dir}main/main_logs/process-$(date +%Y-%m-%d).log
  13. fi
  14. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始监测爬虫进程状态\n" >> ${log_path}
  15. echo "$(date "+%Y-%m-%d %H:%M:%S") 更新环境变量..." >> ${log_path}
  16. cd ~ && source /etc/profile
  17. echo "$(date "+%Y-%m-%d %H:%M:%S") 更新环境变量完成!\n" >> ${log_path}
  18. # 公众号爬虫策略
  19. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 公众号爬虫策略 进程状态" >> ${log_path}
  20. ps -ef | grep "run_gongzhonghao_follow.py" | grep -v "grep"
  21. if [ "$?" -eq 1 ];then
  22. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  23. if [ ${env} = "dev" ];then
  24. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow.py --log_type="follow" --crawler="gongzhonghao" --env="dev" gongzhonghao/nohup-follow.log
  25. else
  26. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow.py --log_type="follow" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow.log
  27. fi
  28. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  29. else
  30. echo "$(date "+%Y-%m-%d %H:%M:%S") 公众号爬虫策略 进程状态正常\n" >> ${log_path}
  31. fi
  32. # 小年糕定向爬虫策略
  33. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕定向爬虫策略 进程状态" >> ${log_path}
  34. ps -ef | grep "run_xiaoniangao_follow.py" | grep -v "grep"
  35. if [ "$?" -eq 1 ];then
  36. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  37. if [ ${env} = "dev" ];then
  38. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="dev" xiaoniangao/nohup-follow.log
  39. else
  40. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-follow.log
  41. fi
  42. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  43. else
  44. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕定向爬虫策略 进程状态正常\n" >> ${log_path}
  45. fi
  46. # 小年糕小时榜爬虫策略
  47. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕小时榜爬虫策略 进程状态" >> ${log_path}
  48. ps -ef | grep "run_xiaoniangao_hour.py" | grep -v "grep"
  49. if [ "$?" -eq 1 ];then
  50. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  51. if [ ${env} = "dev" ];then
  52. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="dev" xiaoniangao/nohup-hour.log
  53. else
  54. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-hour.log
  55. fi
  56. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  57. else
  58. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕小时榜爬虫策略 进程状态正常\n" >> ${log_path}
  59. fi
  60. # 播放量榜爬虫策略
  61. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 播放量榜爬虫策略 进程状态" >> ${log_path}
  62. ps -ef | grep "run_xiaoniangao_play.py" | grep -v "grep"
  63. if [ "$?" -eq 1 ];then
  64. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  65. if [ ${env} = "dev" ];then
  66. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="dev" xiaoniangao/nohup-play.log
  67. else
  68. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-play.log
  69. fi
  70. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  71. else
  72. echo "$(date "+%Y-%m-%d %H:%M:%S") 播放量榜爬虫策略 进程状态正常\n" >> ${log_path}
  73. fi
  74. # 快手定向爬虫策略
  75. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 快手定向爬虫策略 进程状态" >> ${log_path}
  76. ps -ef | grep "run_kuaishou_follow.py" | grep -v "grep"
  77. if [ "$?" -eq 1 ];then
  78. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  79. if [ ${env} = "dev" ];then
  80. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --env="dev" xiaoniangao/nohup-play.log
  81. else
  82. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" kuaishou/follow.log
  83. fi
  84. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  85. else
  86. echo "$(date "+%Y-%m-%d %H:%M:%S") 快手定向爬虫策略 进程状态正常\n" >> ${log_path}
  87. fi
  88. # 西瓜定向爬虫策略
  89. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜定向爬虫策略 进程状态" >> ${log_path}
  90. ps -ef | grep "run_xigua_follow.py" | grep -v "grep"
  91. if [ "$?" -eq 1 ];then
  92. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  93. if [ ${env} = "dev" ];then
  94. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --env="dev" xigua/follow.log
  95. else
  96. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" xigua/follow.log
  97. fi
  98. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  99. else
  100. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜定向爬虫策略 进程状态正常\n" >> ${log_path}
  101. fi
  102. # 删除日志
  103. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始清理 5 天前的日志文件" >> ${log_path}
  104. find ${piaoquan_crawler_dir}main/main_logs/ -mtime +5 -name "*.log" -exec rm -rf {} \;
  105. echo "$(date "+%Y-%m-%d %H:%M:%S") 日志文件清理完毕\n" >> ${log_path}
  106. exit 0
  107. # sh main/process.sh "dev"
  108. # ps aux | grep run_xiaoniangao
  109. # ps aux | grep run_xiaoniangao | grep -v grep | awk '{print $2}' | xargs kill -9
  110. # ps aux | grep run_gongzhonghao
  111. # ps aux | grep run_gongzhonghao | grep -v grep | awk '{print $2}' | xargs kill -9