process.sh 9.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166
  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1-40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1-40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41-81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-2" --crawler="gongzhonghao" --env="dev" gongzhonghao/nohup-follow-2.log
  43. else
  44. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_2.py --log_type="follow-2" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow-2.log
  45. fi
  46. echo "$(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41-81个账号 进程状态正常\n" >> ${log_path}
  49. fi
  50. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 公众号爬虫策略81-121个账号 进程状态" >> ${log_path}
  51. ps -ef | grep "run_gongzhonghao_follow_3.py" | grep -v "grep"
  52. if [ "$?" -eq 1 ];then
  53. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  54. if [ ${env} = "dev" ];then
  55. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_3.py --log_type="follow-3" --crawler="gongzhonghao" --env="dev" gongzhonghao/nohup-follow-3.log
  56. else
  57. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_3.py --log_type="follow-3" --crawler="gongzhonghao" --env="prod" gongzhonghao/nohup-follow-3.log
  58. fi
  59. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  60. else
  61. echo "$(date "+%Y-%m-%d %H:%M:%S") 公众号爬虫策略81-121个账号 进程状态正常\n" >> ${log_path}
  62. fi
  63. # 小年糕定向爬虫策略
  64. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕定向爬虫策略 进程状态" >> ${log_path}
  65. ps -ef | grep "run_xiaoniangao_follow.py" | grep -v "grep"
  66. if [ "$?" -eq 1 ];then
  67. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  68. if [ ${env} = "dev" ];then
  69. 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
  70. else
  71. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-follow.log
  72. fi
  73. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  74. else
  75. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕定向爬虫策略 进程状态正常\n" >> ${log_path}
  76. fi
  77. # 小年糕小时榜爬虫策略
  78. echo "$(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" >> ${log_path}
  79. ps -ef | grep "run_xiaoniangao_hour.py" | grep -v "grep"
  80. if [ "$?" -eq 1 ];then
  81. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  82. if [ ${env} = "dev" ];then
  83. 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
  84. else
  85. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-hour.log
  86. fi
  87. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  88. else
  89. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕小时榜爬虫策略 进程状态正常\n" >> ${log_path}
  90. fi
  91. # 播放量榜爬虫策略
  92. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 播放量榜爬虫策略 进程状态" >> ${log_path}
  93. ps -ef | grep "run_xiaoniangao_play.py" | grep -v "grep"
  94. if [ "$?" -eq 1 ];then
  95. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  96. if [ ${env} = "dev" ];then
  97. 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
  98. else
  99. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="prod" xiaoniangao/nohup-play.log
  100. fi
  101. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  102. else
  103. echo "$(date "+%Y-%m-%d %H:%M:%S") 播放量榜爬虫策略 进程状态正常\n" >> ${log_path}
  104. fi
  105. # 快手定向爬虫策略
  106. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 快手定向爬虫策略 进程状态" >> ${log_path}
  107. ps -ef | grep "run_kuaishou_follow.py" | grep -v "grep"
  108. if [ "$?" -eq 1 ];then
  109. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  110. if [ ${env} = "dev" ];then
  111. 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
  112. else
  113. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" kuaishou/follow.log
  114. fi
  115. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  116. else
  117. echo "$(date "+%Y-%m-%d %H:%M:%S") 快手定向爬虫策略 进程状态正常\n" >> ${log_path}
  118. fi
  119. # 西瓜定向爬虫策略
  120. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜定向爬虫策略 进程状态" >> ${log_path}
  121. ps -ef | grep "run_xigua_follow.py" | grep -v "grep"
  122. if [ "$?" -eq 1 ];then
  123. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  124. if [ ${env} = "dev" ];then
  125. 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
  126. else
  127. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" xigua/follow.log
  128. fi
  129. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  130. else
  131. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜定向爬虫策略 进程状态正常\n" >> ${log_path}
  132. fi
  133. # youtube定向爬虫策略
  134. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 youtube定向爬虫策略 进程状态" >> ${log_path}
  135. ps -ef | grep "run_youtube_follow.py" | grep -v "grep"
  136. if [ "$?" -eq 1 ];then
  137. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  138. if [ ${env} = "dev" ];then
  139. 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
  140. else
  141. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./youtube/youtube_main/run_youtube_follow.py --log_type="follow" --crawler="youtube" --strategy="定向爬虫策略" --oss_endpoint="hk" --env="hk" --machine="aliyun_hk" youtube/follow.log
  142. fi
  143. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!\n" >> ${log_path}
  144. else
  145. echo "$(date "+%Y-%m-%d %H:%M:%S") youtube定向爬虫策略 进程状态正常\n" >> ${log_path}
  146. fi
  147. # 删除日志
  148. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始清理 5 天前的日志文件" >> ${log_path}
  149. find ${piaoquan_crawler_dir}main/main_logs/ -mtime +5 -name "*.log" -exec rm -rf {} \;
  150. echo "$(date "+%Y-%m-%d %H:%M:%S") 日志文件清理完毕\n" >> ${log_path}
  151. exit 0