process.sh 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130
  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1-50个账号 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1-50个账号 进程状态正常\n" >> ${log_path}
  31. fi
  32. #echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 公众号爬虫策略51-100个账号 进程状态" >> ${log_path}
  33. #ps -ef | grep "run_gongzhonghao_follow_2.py" | grep -v "grep"
  34. #if [ "$?" -eq 1 ];then
  35. # echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  36. # if [ ${env} = "dev" ];then
  37. # cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_2.py --log_type="follow" --crawler="gongzhonghao" --env="dev" gongzhonghao/nohup-follow.log
  38. # else
  39. # cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_2.py --log_type="follow" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow-2.log
  40. # fi
  41. # echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  42. #else
  43. # echo "$(date "+%Y-%m-%d %H:%M:%S") 公众号爬虫策略1-50个账号 进程状态正常\n" >> ${log_path}
  44. #fi
  45. # 小年糕定向爬虫策略
  46. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕定向爬虫策略 进程状态" >> ${log_path}
  47. ps -ef | grep "run_xiaoniangao_follow.py" | grep -v "grep"
  48. if [ "$?" -eq 1 ];then
  49. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  50. if [ ${env} = "dev" ];then
  51. 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
  52. else
  53. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-follow.log
  54. fi
  55. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  56. else
  57. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕定向爬虫策略 进程状态正常\n" >> ${log_path}
  58. fi
  59. # 小年糕小时榜爬虫策略
  60. echo "$(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" >> ${log_path}
  61. ps -ef | grep "run_xiaoniangao_hour.py" | grep -v "grep"
  62. if [ "$?" -eq 1 ];then
  63. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  64. if [ ${env} = "dev" ];then
  65. 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
  66. else
  67. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-hour.log
  68. fi
  69. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  70. else
  71. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕小时榜爬虫策略 进程状态正常\n" >> ${log_path}
  72. fi
  73. # 播放量榜爬虫策略
  74. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 播放量榜爬虫策略 进程状态" >> ${log_path}
  75. ps -ef | grep "run_xiaoniangao_play.py" | grep -v "grep"
  76. if [ "$?" -eq 1 ];then
  77. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  78. if [ ${env} = "dev" ];then
  79. 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
  80. else
  81. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-play.log
  82. fi
  83. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  84. else
  85. echo "$(date "+%Y-%m-%d %H:%M:%S") 播放量榜爬虫策略 进程状态正常\n" >> ${log_path}
  86. fi
  87. ## 快手定向爬虫策略
  88. #echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 快手定向爬虫策略 进程状态" >> ${log_path}
  89. #ps -ef | grep "run_kuaishou_follow.py" | grep -v "grep"
  90. #if [ "$?" -eq 1 ];then
  91. # echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  92. # if [ ${env} = "dev" ];then
  93. # 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
  94. # else
  95. # cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" kuaishou/follow.log
  96. # fi
  97. # echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  98. #else
  99. # echo "$(date "+%Y-%m-%d %H:%M:%S") 快手定向爬虫策略 进程状态正常\n" >> ${log_path}
  100. #fi
  101. # 西瓜定向爬虫策略
  102. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜定向爬虫策略 进程状态" >> ${log_path}
  103. ps -ef | grep "run_xigua_follow.py" | grep -v "grep"
  104. if [ "$?" -eq 1 ];then
  105. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  106. if [ ${env} = "dev" ];then
  107. 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
  108. else
  109. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" xigua/follow.log
  110. fi
  111. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  112. else
  113. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜定向爬虫策略 进程状态正常\n" >> ${log_path}
  114. fi
  115. # 删除日志
  116. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始清理 5 天前的日志文件" >> ${log_path}
  117. find ${piaoquan_crawler_dir}main/main_logs/ -mtime +5 -name "*.log" -exec rm -rf {} \;
  118. echo "$(date "+%Y-%m-%d %H:%M:%S") 日志文件清理完毕\n" >> ${log_path}
  119. exit 0