process.sh 8.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152
  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. elif [ ${env} = "hk" ];then
  9. piaoquan_crawler_dir=/root/piaoquan_crawler/
  10. profile_path=/etc/profile
  11. python=python3
  12. log_path=${piaoquan_crawler_dir}main/main_logs/process-$(date +%Y-%m-%d).log
  13. else
  14. piaoquan_crawler_dir=/data5/piaoquan_crawler/
  15. profile_path=/etc/profile
  16. python=python
  17. log_path=${piaoquan_crawler_dir}main/main_logs/process-$(date +%Y-%m-%d).log
  18. fi
  19. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始监测爬虫进程状态\n" >> ${log_path}
  20. echo "$(date "+%Y-%m-%d %H:%M:%S") 更新环境变量..." >> ${log_path}
  21. cd ~ && source /etc/profile
  22. echo "$(date "+%Y-%m-%d %H:%M:%S") 更新环境变量完成!\n" >> ${log_path}
  23. # 公众号爬虫策略
  24. echo "$(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" >> ${log_path}
  25. ps -ef | grep "run_gongzhonghao_follow.py" | grep -v "grep"
  26. if [ "$?" -eq 1 ];then
  27. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  28. if [ ${env} = "dev" ];then
  29. 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
  30. else
  31. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow.py --log_type="follow" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow.log
  32. fi
  33. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  34. else
  35. echo "$(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" >> ${log_path}
  36. fi
  37. #echo "$(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" >> ${log_path}
  38. #ps -ef | grep "run_gongzhonghao_follow_2.py" | grep -v "grep"
  39. #if [ "$?" -eq 1 ];then
  40. # echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  41. # if [ ${env} = "dev" ];then
  42. # 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
  43. # else
  44. # cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_2.py --log_type="follow" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow-2.log
  45. # fi
  46. # echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  47. #else
  48. # echo "$(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" >> ${log_path}
  49. #fi
  50. # 小年糕定向爬虫策略
  51. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕定向爬虫策略 进程状态" >> ${log_path}
  52. ps -ef | grep "run_xiaoniangao_follow.py" | grep -v "grep"
  53. if [ "$?" -eq 1 ];then
  54. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  55. if [ ${env} = "dev" ];then
  56. 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
  57. else
  58. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-follow.log
  59. fi
  60. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  61. else
  62. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕定向爬虫策略 进程状态正常\n" >> ${log_path}
  63. fi
  64. # 小年糕小时榜爬虫策略
  65. echo "$(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" >> ${log_path}
  66. ps -ef | grep "run_xiaoniangao_hour.py" | grep -v "grep"
  67. if [ "$?" -eq 1 ];then
  68. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  69. if [ ${env} = "dev" ];then
  70. 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
  71. else
  72. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-hour.log
  73. fi
  74. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  75. else
  76. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕小时榜爬虫策略 进程状态正常\n" >> ${log_path}
  77. fi
  78. # 播放量榜爬虫策略
  79. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 播放量榜爬虫策略 进程状态" >> ${log_path}
  80. ps -ef | grep "run_xiaoniangao_play.py" | grep -v "grep"
  81. if [ "$?" -eq 1 ];then
  82. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  83. if [ ${env} = "dev" ];then
  84. 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
  85. else
  86. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-play.log
  87. fi
  88. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  89. else
  90. echo "$(date "+%Y-%m-%d %H:%M:%S") 播放量榜爬虫策略 进程状态正常\n" >> ${log_path}
  91. fi
  92. # 快手定向爬虫策略
  93. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 快手定向爬虫策略 进程状态" >> ${log_path}
  94. ps -ef | grep "run_kuaishou_follow.py" | grep -v "grep"
  95. if [ "$?" -eq 1 ];then
  96. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  97. if [ ${env} = "dev" ];then
  98. 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
  99. else
  100. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" kuaishou/follow.log
  101. fi
  102. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  103. else
  104. echo "$(date "+%Y-%m-%d %H:%M:%S") 快手定向爬虫策略 进程状态正常\n" >> ${log_path}
  105. fi
  106. # 西瓜定向爬虫策略
  107. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜定向爬虫策略 进程状态" >> ${log_path}
  108. ps -ef | grep "run_xigua_follow.py" | grep -v "grep"
  109. if [ "$?" -eq 1 ];then
  110. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  111. if [ ${env} = "dev" ];then
  112. 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
  113. else
  114. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" xigua/follow.log
  115. fi
  116. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  117. else
  118. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜定向爬虫策略 进程状态正常\n" >> ${log_path}
  119. fi
  120. # youtube定向爬虫策略
  121. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 youtube定向爬虫策略 进程状态" >> ${log_path}
  122. ps -ef | grep "run_youtube_follow.py" | grep -v "grep"
  123. if [ "$?" -eq 1 ];then
  124. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  125. if [ ${env} = "dev" ];then
  126. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./youtube/youtube_main/run_youtube_follow.py --log_type="follow" --crawler="youtube" --env="dev" youtube/follow.log
  127. else
  128. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./youtube/youtube_main/run_youtube_follow.py --log_type="follow" --crawler="youtube" --strategy="定向爬虫策略" --oss_endpoint="hk" --env="prod" --machine="aliyun_hk" youtube/follow.log
  129. fi
  130. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  131. else
  132. echo "$(date "+%Y-%m-%d %H:%M:%S") youtube定向爬虫策略 进程状态正常\n" >> ${log_path}
  133. fi
  134. # 删除日志
  135. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始清理 5 天前的日志文件" >> ${log_path}
  136. find ${piaoquan_crawler_dir}main/main_logs/ -mtime +5 -name "*.log" -exec rm -rf {} \;
  137. echo "$(date "+%Y-%m-%d %H:%M:%S") 日志文件清理完毕\n" >> ${log_path}
  138. exit 0