postgresql 监控 session block tree

session 阻塞的监控在数据库是重要的监控指标之一。
参考
https://wiki.postgresql.org/wiki/Lock_Monitoring

12、11、10、9.6

with recursive tmp_lock as (
    select distinct
           --w.mode w_mode,w.page w_page,
           --w.tuple w_tuple,w.xact_start w_xact_start,w.query_start w_query_start,
           --now()-w.query_start w_locktime,w.query w_query 
           w.pid as id,--w_pid,
           r.pid as parentid--r_pid,
           --r.locktype,r.mode r_mode,r.usename r_user,r.datname r_db,
           --r.relation::regclass,
           --r.page r_page,r.tuple r_tuple,r.xact_start r_xact_start,
           --r.query_start r_query_start,
           --now()-r.query_start r_locktime,r.query r_query,   
    from (
          select a.mode,a.locktype,a.database,
                 a.relation,a.page,a.tuple,a.classid,
                 a.objid,a.objsubid,a.pid,a.virtualtransaction,a.virtualxid,
                 a.transactionid,
                 b.query as query,
                 b.xact_start,b.query_start,b.usename,b.datname 
           from pg_locks a,
                pg_stat_activity b 
          where a.pid=b.pid 
            and not a.granted 
          ) w,
         (
          select a.mode,a.locktype,a.database,
                 a.relation,a.page,a.tuple,a.classid,
                 a.objid,a.objsubid,a.pid,a.virtualtransaction,a.virtualxid,
                 a.transactionid,
                 b.query as query,
                 b.xact_start,b.query_start,b.usename,b.datname 
            from pg_locks a,
                 pg_stat_activity b -- select pg_typeof(pid) from pg_stat_activity
           where a.pid=b.pid 
             and a.granted 
          ) r 
    where 1=1
      and r.locktype is not distinct from w.locktype 
      and r.database is not distinct from w.database 
      and r.relation is not distinct from w.relation 
      and r.page is not distinct from w.page 
      and r.tuple is not distinct from w.tuple 
      and r.classid is not distinct from w.classid 
      and r.objid is not distinct from w.objid 
      and r.objsubid is not distinct from w.objsubid 
      and r.transactionid is not distinct from w.transactionid 
      and r.pid <> w.pid
),tmp0 as (
  select *
    from tmp_lock tl
   union all
  --查找root，同一时刻可能有多个root
  select t1.parentid,0::int4
    from tmp_lock t1
   where 1=1
     and t1.parentid not in (select id from tmp_lock)
),tmp3 (pathid,depth,id,parentid) as (
	--对过滤出的机构向下递归，构成tree
	SELECT array[id]::text[] as pathid,1 as depth,id,parentid
	  FROM tmp0
	 where 1=1
	   and parentid=0
	 union
	SELECT t0.pathid||array[t1.id]::text[] as pathid,t0.depth+1 as depth,t1.id,t1.parentid
	  FROM tmp0 t1,  
	       tmp3 t0
	 where 1=1
	   and t1.parentid=t0.id
)
select distinct 
       '/'||array_to_string(a0.pathid,'/') as pathid,
       a0.depth,
       a0.id,a0.parentid,lpad(a0.id::text, 2*a0.depth-1+length(a0.id::text),' ') as tree_id,
       --'select pg_cancel_backend('||a0.id|| ');' as cancel_pid,
       --'select pg_terminate_backend('||a0.id|| ');' as term_pid,
       case when a0.depth =1 then 'select pg_terminate_backend('|| a0.id || ');' else null end  as term_pid,
       case when a0.depth =1 then 'select cancel_backend('|| a0.id || ');' else null end  as cancel_pid
       ,a2.datname,a2.usename,a2.application_name,a2.client_addr,a2.wait_event_type,a2.wait_event,a2.state
       --,a2.backend_start,a2.xact_start,a2.query_start
  from tmp3 a0
       left outer join (select distinct '/'||id||'/' as prefix_id,id
	                      from tmp0
						 where 1=1 ) a1
	                on position( a1.prefix_id in '/'||array_to_string(a0.pathid,'/')||'/' ) >0
	   left outer join pg_stat_activity a2 -- select * from pg_stat_activity
	                on a0.id = a2.pid
order by '/'||array_to_string(a0.pathid,'/'),a0.depth  
;

结果如下

       pathid       | depth |  id   | parentid |  tree_id   | datname  | usename  | application_name | client_addr | wait_event_type |  wait_event   |        state        |         backend_start         |          xact_start           |          query_start          
--------------------+-------+-------+----------+------------+----------+----------+------------------+-------------+-----------------+---------------+---------------------+-------------------------------+-------------------------------+-------------------------------
 /10180             |     1 | 10180 |        0 |  10180     | postgres | postgres | psql             |             |                 |               | idle in transaction | 2018-11-05 15:25:56.473886+08 | 2018-11-05 15:26:12.529745+08 | 2018-11-05 16:30:50.347912+08
 /10180/10197       |     2 | 10197 |    10180 |    10197   | postgres | postgres | psql             |             | Lock            | transactionid | active              | 2018-11-05 15:26:02.12397+08  | 2018-11-05 15:26:30.767745+08 | 2018-11-05 15:26:30.767745+08
 /10180/10197/10508 |     3 | 10508 |    10197 |      10508 | postgres | postgres | psql             |             | Lock            | tuple         | active              | 2018-11-05 16:38:02.858333+08 | 2018-11-05 16:38:19.514655+08 | 2018-11-05 16:38:19.514655+08
 /9782              |     1 |  9782 |        0 |  9782      | postgres | postgres | psql             |             |                 |               | idle in transaction | 2018-11-05 15:05:04.656155+08 | 2018-11-05 15:11:45.731423+08 | 2018-11-05 15:46:24.559188+08
 /9782/10026        |     2 | 10026 |     9782 |    10026   | postgres | postgres | psql             |             | Lock            | transactionid | active              | 2018-11-05 15:24:02.389744+08 | 2018-11-05 15:50:16.546751+08 | 2018-11-05 15:50:38.428968+08
 /9782/10026/9918   |     3 |  9918 |    10026 |      9918  | postgres | postgres | psql             |             | Lock            | tuple         | active              | 2018-11-05 15:11:31.394268+08 | 2018-11-05 15:50:19.964021+08 | 2018-11-05 15:50:45.080284+08
(6 rows)