process.sh 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197
  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" >> ${log_path}
  20. echo "$(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!" >> ${log_path}
  23. # 公众号爬虫策略
  24. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 公众号爬虫策略 1-120个账号 进程状态" >> ${log_path}
  25. #echo "$(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" >> ${log_path}
  26. ps -ef | grep "run_gongzhonghao_follow.py" | grep -v "grep"
  27. if [ "$?" -eq 1 ];then
  28. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  29. if [ ${env} = "dev" ];then
  30. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow.py --log_type="follow" --crawler="gongzhonghao" --env="dev" gongzhonghao/logs/nohup-follow.log
  31. else
  32. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow.py --log_type="follow" --crawler="gongzhonghao" --env="prod" gongzhonghao/logs/nohup-follow.log
  33. fi
  34. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  35. else
  36. echo "$(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" >> ${log_path}
  37. fi
  38. #echo "$(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" >> ${log_path}
  39. #ps -ef | grep "run_gongzhonghao_follow_2.py" | grep -v "grep"
  40. #if [ "$?" -eq 1 ];then
  41. # echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  42. # if [ ${env} = "dev" ];then
  43. # 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/logs/nohup-follow-2.log
  44. # else
  45. # cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_2.py --log_type="follow-2" --crawler="gongzhonghao" --env="prod" gongzhonghao/logs/nohup-follow-2.log
  46. # fi
  47. # echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  48. #else
  49. # echo "$(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" >> ${log_path}
  50. #fi
  51. #
  52. #echo "$(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" >> ${log_path}
  53. #ps -ef | grep "run_gongzhonghao_follow_3.py" | grep -v "grep"
  54. #if [ "$?" -eq 1 ];then
  55. # echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  56. # if [ ${env} = "dev" ];then
  57. # 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/logs/nohup-follow-3.log
  58. # else
  59. # cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./gongzhonghao/gongzhonghao_main/run_gongzhonghao_follow_3.py --log_type="follow-3" --crawler="gongzhonghao" --env="prod" gongzhonghao/logs/nohup-follow-3.log
  60. # fi
  61. # echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  62. #else
  63. # echo "$(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" >> ${log_path}
  64. #fi
  65. # 小年糕定向爬虫策略
  66. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 小年糕定向爬虫策略 进程状态" >> ${log_path}
  67. ps -ef | grep "run_xiaoniangao_follow.py" | grep -v "grep"
  68. if [ "$?" -eq 1 ];then
  69. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  70. if [ ${env} = "dev" ];then
  71. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="dev" xiaoniangao/logs/nohup-follow.log
  72. else
  73. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_follow.py --log_type="follow" --crawler="xiaoniangao" --env="prod" xiaoniangao/logs/nohup-follow.log
  74. fi
  75. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  76. else
  77. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕定向爬虫策略 进程状态正常" >> ${log_path}
  78. fi
  79. # 小年糕小时榜爬虫策略
  80. echo "$(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" >> ${log_path}
  81. ps -ef | grep "run_xiaoniangao_hour.py" | grep -v "grep"
  82. if [ "$?" -eq 1 ];then
  83. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  84. if [ ${env} = "dev" ];then
  85. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="dev" xiaoniangao/logs/nohup-hour.log
  86. else
  87. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_hour.py --log_type="hour" --crawler="xiaoniangao" --env="prod" xiaoniangao/logs/nohup-hour.log
  88. fi
  89. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  90. else
  91. echo "$(date "+%Y-%m-%d %H:%M:%S") 小年糕小时榜爬虫策略 进程状态正常" >> ${log_path}
  92. fi
  93. # 播放量榜爬虫策略
  94. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 播放量榜爬虫策略 进程状态" >> ${log_path}
  95. ps -ef | grep "run_xiaoniangao_play.py" | grep -v "grep"
  96. if [ "$?" -eq 1 ];then
  97. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  98. if [ ${env} = "dev" ];then
  99. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="dev" xiaoniangao/logs/nohup-play.log
  100. else
  101. cd ${piaoquan_crawler_dir} && /usr/bin/sh /data5/piaoquan_crawler/main/scheduling_main.sh ./xiaoniangao/xiaoniangao_main/run_xiaoniangao_play.py --log_type="play" --crawler="xiaoniangao" --env="prod" xiaoniangao/logs/nohup-play.log
  102. fi
  103. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  104. else
  105. echo "$(date "+%Y-%m-%d %H:%M:%S") 播放量榜爬虫策略 进程状态正常" >> ${log_path}
  106. fi
  107. # 快手定向爬虫策略
  108. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 快手定向爬虫策略 进程状态" >> ${log_path}
  109. ps -ef | grep "run_kuaishou_follow.py" | grep -v "grep"
  110. if [ "$?" -eq 1 ];then
  111. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  112. if [ ${env} = "dev" ];then
  113. 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
  114. else
  115. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./kuaishou/kuaishou_main/run_kuaishou_follow.py --log_type="follow" --crawler="kuaishou" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" kuaishou/follow.log
  116. fi
  117. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  118. else
  119. echo "$(date "+%Y-%m-%d %H:%M:%S") 快手定向爬虫策略 进程状态正常" >> ${log_path}
  120. fi
  121. #
  122. # 抖音推荐爬虫策略
  123. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 抖音推荐爬虫策略 进程状态" >> ${log_path}
  124. ps -ef | grep "run_douyin_recommend.py" | grep -v "grep"
  125. if [ "$?" -eq 1 ];then
  126. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  127. if [ ${env} = "dev" ];then
  128. 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
  129. else
  130. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./douyin/douyin_main/run_douyin_recommend.py --log_type="recommend" --crawler="douyin" --strategy="抖音推荐爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" douyin/recommend.log
  131. fi
  132. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  133. else
  134. echo "$(date "+%Y-%m-%d %H:%M:%S") 抖音推荐爬虫策略 进程状态正常" >> ${log_path}
  135. fi
  136. # 西瓜定向爬虫策略
  137. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜定向爬虫策略 进程状态" >> ${log_path}
  138. ps -ef | grep "run_xigua_follow.py" | grep -v "grep"
  139. if [ "$?" -eq 1 ];then
  140. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  141. if [ ${env} = "dev" ];then
  142. 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
  143. else
  144. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./xigua/xigua_main/run_xigua_follow.py --log_type="follow" --crawler="xigua" --strategy="定向爬虫策略" --oss_endpoint="inner" --env="prod" --machine="aliyun" xigua/follow.log
  145. fi
  146. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  147. else
  148. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜定向爬虫策略 进程状态正常" >> ${log_path}
  149. fi
  150. # 西瓜推荐榜爬虫策略
  151. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 西瓜推荐榜爬虫策略 进程状态" >> ${log_path}
  152. ps -ef | grep "run_xigua_recommend.py" | grep -v "grep"
  153. if [ "$?" -eq 1 ];then
  154. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  155. if [ ${env} = "dev" ];then
  156. cd ${piaoquan_crawler_dir} && sh main/scheduling_main.sh ./xigua/xigua_main/run_xigua_recommend.py --log_type="recommend" --crawler="xigua" --env="dev" xigua/logs/nohup-recommend.log
  157. else
  158. cd ${piaoquan_crawler_dir} && /usr/bin/sh main/scheduling_main.sh ./xigua/xigua_main/run_xigua_recommend.py --log_type="recommend" --crawler="xigua" --env="prod" xigua/logs/nohup-recommend.log
  159. fi
  160. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  161. else
  162. echo "$(date "+%Y-%m-%d %H:%M:%S") 西瓜推荐榜爬虫策略 进程状态正常" >> ${log_path}
  163. fi
  164. # youtube定向爬虫策略
  165. echo "$(date "+%Y-%m-%d %H:%M:%S") 正在监测 youtube定向爬虫策略 进程状态" >> ${log_path}
  166. ps -ef | grep "run_youtube_follow.py" | grep -v "grep"
  167. if [ "$?" -eq 1 ];then
  168. echo "$(date "+%Y-%m-%d_%H:%M:%S") 异常停止,正在重启!" >> ${log_path}
  169. if [ ${env} = "dev" ];then
  170. 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
  171. else
  172. cd ${piaoquan_crawler_dir} && /usr/bin/sh ./main/main.sh ./youtube/youtube_main/run_youtube_follow.py --log_type="follow" --crawler="youtube" --strategy="定向爬虫策略" --oss_endpoint="hk" --env="hk" --machine="aliyun_hk" youtube/follow.log
  173. fi
  174. echo "$(date "+%Y-%m-%d %H:%M:%S") 重启完成!" >> ${log_path}
  175. else
  176. echo "$(date "+%Y-%m-%d %H:%M:%S") youtube定向爬虫策略 进程状态正常" >> ${log_path}
  177. fi
  178. # 删除日志
  179. echo "$(date "+%Y-%m-%d %H:%M:%S") 开始清理 5 天前的日志文件" >> ${log_path}
  180. find ${piaoquan_crawler_dir}main/main_logs/ -mtime +5 -name "*.log" -exec rm -rf {} \;
  181. echo "$(date "+%Y-%m-%d %H:%M:%S") 日志文件清理完毕" >> ${log_path}
  182. exit 0