PostgreSQL 流複製監控腳本

https://github.com/AndyYHM/Writing/blob/PostgreSQL/20181219-PostgreSQL%20Stream%20MON.md

整理之前的腳本文檔，搬家至博客園，梳理一下之前寫的shell腳本

適用於PostgreSQL版本10、版本9替換函數名稱即可，系統centos 7驗證

_xlog_location<=> _wal_lsn

_location <=> _lsn

1. 腳本輸出效果

1.1 主節點顯示效果

        STREAM_ROLE    : Master         
        ------------------------------
        Replication Client Info:
        ------------------------------
        PID            : 25424          
        CLIENT_ADDR    : 20.9.6.213     
        SYNC_STATE     : async          
        STATE          : streaming      
        WRITE_DIFF     : 0bytes         
        FLUSH_DIFF     : 0bytes         
        REPLAY_DIFF    : 0bytes         

        Replication Client Info:
        ------------------------------
        PID            : 24586          
        CLIENT_ADDR    : 20.9.6.212     
        SYNC_STATE     : sync           
        STATE          : streaming      
        WRITE_DIFF     : 0bytes         
        FLUSH_DIFF     : 0bytes         
        REPLAY_DIFF    : 0bytes

1.2 從節點效果

STREAM_ROLE    : Slave          
------------------------------
Replication Info
------------------------------
CLUSTER_STATE  : in_archive_recovery
SERVER_ADDR    : 20.9.6.219     
USER           : replicator     
APP_NAME       : node12         
TARGET_TIMELINE:                
READ_ONLY      : on  

2. 功能說明

• 判定主機角色
• 如果為Master節點，則顯示覆制節點信息，及與主節點差值
• 如果為Replication節點，則顯示為本節點相關信息

3. 使用說明

• 根據實際環境修改腳本pgdata=""
• postgres系統用戶執行，保證有執行許可權
• 本地執行腳本sh pg10_stream.sh

4. pg10_stream.sh腳本內容

#!/bin/bash
################################
## Author: [email protected]
## Version: 1.0
## Date:    20181219
################################
clear

host_var="/tmp"

pgdata="/pgdb/pgdata"

work_dir=$(cd `dirname $0`;pwd)

cd $work_dir

is_master=$(psql -h $host_var -c 'select  pg_is_in_recovery();'| sed -n 3p)


if [ "$is_master" == " f" ];then

    rep=$(psql  -h $host_var -c 'select  pid, client_addr, sync_state, state,pg_size_pretty( pg_wal_lsn_diff(pg_current_wal_lsn(),write_lsn)) as write_diff, pg_size_pretty(pg_wal_lsn_diff(pg_current_wal_lsn(),flush_lsn)) as flush_diff, pg_size_pretty(pg_wal_lsn_diff(pg_current_wal_lsn(),replay_lsn)) as replay_diff      from pg_stat_replication;'| sed -n '3,$'p |tac  | sed -n '3,$'p | sed s/" *"//g )
    if [ ! "$rep" ];then
      echo "Waring! No Replication Client Found"
    else
      printf "\n\t%-15s: %-15s\n" "STREAM_ROLE" "Master"
      echo  -e "\t------------------------------"

      echo "$rep" | while read i
        do
          echo  -e "\tReplication Client Info:\n\t------------------------------"
          a=(null PID CLIENT_ADDR SYNC_STATE STATE WRITE_DIFF FLUSH_DIFF REPLAY_DIFF)
          for j in `seq 7`
            do
              k=${a[j]}
              echo -e "$i" | awk -v k=$k -v j=$j -F"|" '{printf "\t%-15s: %-15s\n",k,$j}'
            done
          echo ""
        done


      client_check_info=(`psql -h $host_var -c 'select client_addr from pg_stat_replication;' | sed -n '3,$'p |tac  | sed -n '3,$'p | sed s/" *"//g`)
      if [ ! -f pg_stream_client ]; then
        echo "$client_check_info" > pg_stream_client
      else
        client_before=(`cat pg_stream_client`)
        if [ ${#client_before[*]} -gt 0 ]; then
          for i in `seq ${#client_before[*]}`
            do
              index=$(expr $i - 1)
              if [ "${client_check_info[$index]}"x !=  "${client_before[$index]}"x ];then
                printf  "\tWarning! Replication Client Lost.\n"
              else
                echo "$client_check_info" > pg_stream_client

              fi
            done
         else
           printf  "\tAttention! No Client Info found in pg_stream_client.\n"
         fi
      fi
    fi


elif [ "$is_master" == " t" ];then


    declare -A connect_info_dict

    conninfo=$(cat $pgdata/recovery.conf| grep -i "^primary_conninfo" | awk -F"'" '{print $2}' |sed s/"  *"/" "/g |sed s/"="/"\"]=\""/g | sed s/" "/"\" [\""/g | sed s/^/"[\""/g |sed s/$/"\""/g|sed s/" "/"\n"/g)


    eval "connect_info_dict=($conninfo)"

    cluster_state=$(pg_controldata  | grep cluster | sed s/"  *"/" "/g | cut -d ":" -f2| sed s/"^ "//g|  sed s/" "/"_"/g)
    server_addr=${connect_info_dict["host"]}
    user=${connect_info_dict["user"]}
    application_name=${connect_info_dict["application_name"]}
    recovery_target_timeline=${connect_info_dict["recovery_target_timeline"]}
    read_only=$(psql  -h $host_var -c 'show  transaction_read_only;' | sed -n '3p')


    output(){
      printf "\t%-15s: %-15s\n" $1 $2
    }



    printf "\n\t%-15s: %-15s\n" "STREAM_ROLE" "Slave"
    echo  -e "\t------------------------------"
    echo  -e "\tReplication Info\n\t------------------------------"
    output "CLUSTER_STATE" $cluster_state
    output "SERVER_ADDR" $server_addr
    output "USER" $user
    output "APP_NAME" $application_name
    output "TARGET_TIMELINE" $recovery_target_timeline
    output "READ_ONLY" $read_only
    echo ""

else

    echo "Warning! Can't Connect to Database. Please Check Database Status "
fi

exit 0

psql  -c 'select application_name,client_addr ,sync_state from pg_stat_replication;' | sed -n '3,$'p |tac  | sed -n '3,$'p | sed s/" *"//g