今天主要介绍Oracle数据库在并行操作过程中 slave 进程和 QC 进程经常遇到的等待事件及常用脚本。
一、PX Deq: Execution Msg，PX Deq: Execute Reply等待事件
1. PX Deq: Execution Msg
| Occurs when a parallel slave is waiting to be told what to do. This is normally considered an idle event, but can cause excessive CPU in some cases. |
该事件是并行查询中的常见事件。当PQ slave进程在等待QC告诉它要做什么的时候就会出现此事件(eg: when waiting to be told parse / execute / fetch etc..)
v$session_wait 中该等待事件对应的参数：
- P1 = sleeptime/senderid
- P2 = passes
- P3 = not used
我们可以使用如下语句获取转换sleeptime/senderid的相关信息：
set SERVEROUTPUT on
undef p1
declare
inst varchar(20);
sender varchar(20);
begin
select bitand(&&p1, 16711680) - 65535 as SNDRINST,
decode(bitand(&&p1, 65535),65535, 'QC', 'P'||to_char(bitand(&&p1, 65535),'fm000') ) as SNDR
into inst , sender
from dual
where bitand(&&p1, 268435456) = 268435456;
dbms_output.put_line('Instance = '||inst);
dbms_output.put_line('Sender = '||sender );
end;
/
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
如果P1的值为空，则意味slave 不需要等待任何进程
比如p1的值为268501004，则上面的sql会返回：
Instance = 1
Sender = P012
- 1.
- 2.
passes 进程在得到信息之前循环轮转等待的次数
该等待事件是一个空闲等待事件，当此等待事件出现,进程会持续等待并逐渐增加等待次数直到获取信息!
解决方法：
作为 Coordinator 的 Process 在获取 Slave 进程的数据时，反应太慢了，导致某些 Slave进行因为 Queue 满而不得不等待，进而拖慢了整个并行执行的速度。
这常常是由于 CPU 数目不足或者 系统中运行的 进程太多导致。可考虑 减小并行度。
2. PX Deq: Execute Reply
| Occurs when the query coordinator is waiting for a response from a parallel slave. This is normally considered an idle event, but can cause excessive CPU in some cases.
Waiting Process: QC |
协调器正在等待一个 从slaves 进程对控制信息的响应(确认通知)或者期望从slave进程集中获取数据。这个等待事件意味着QC等待slaves结束执行sql 并且将结果集发送给QC
v$session_wait 中该等待事件对应的参数：
- P1 = sleeptime/senderid
- P2 = passes
- P3 = not used
我们可以使用如下语句获取转换sleeptime/senderid的相关信息：
set SERVEROUTPUT on
undef p1
declare
inst varchar(20);
sender varchar(20);
begin
select bitand(&&p1, 16711680) - 65535 as SNDRINST,
decode(bitand(&&p1, 65535),65535, 'QC', 'P'||to_char(bitand(&&p1, 65535),'fm000') ) as SNDR
into inst , sender
from dual
where bitand(&&p1, 268435456) = 268435456;
dbms_output.put_line('Instance = '||inst);
dbms_output.put_line('Sender = '||sender );
end;
/
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
如果P1的值为空，则意味slave 不需要等待任何进程
比如p1的值为268501004，则上面的sql会返回：
Instance = 1
Sender = P012
- 1.
- 2.
等待时间：这是非空闲等待时间，QC 等待从slave 的响应或者查询的数据结果
解决办法：非优化的sql语句肯能是导致此等待事件的原因：slaves 需要花费很长时间来执行sql 语句而qc又在等待slave返回数据。
优化sql，查看slave 在执行的语句以及其执行计划，并做出尽量的优化,以便减少slave执行sql语句的时间!
二、相关脚本
1. gives an overview of all running parallel queries with all slaves.It shows the if a slave is waiting and for what event it waits.
select decode(px.qcinst_id,
NULL,
username,
' - ' ||
lower(substr(pp.SERVER_NAME, length(pp.SERVER_NAME) - 4, 4))) "Username",
decode(px.qcinst_id, NULL, 'QC', '(Slave)') "QC/Slave",
to_char(px.server_set) "SlaveSet",
to_char(s.sid) "SID",
to_char(px.inst_id) "Slave INST",
decode(sw.state, 'WAITING', 'WAIT', 'NOT WAIT') as STATE,
case sw.state
WHEN 'WAITING' THEN
substr(sw.event, 1, 30)
ELSE
NULL
end as wait_event,
decode(px.qcinst_id, NULL, to_char(s.sid), px.qcsid) "QC SID",
to_char(px.qcinst_id) "QC INST",
px.req_degree "Req. DOP",
px.degree "Actual DOP"
from gv$px_session px, gv$session s, gv$px_process pp, gv$session_wait sw
where px.sid = s.sid(+)
and px.serial# = s.serial#(+)
and px.inst_id = s.inst_id(+)
and px.sid = pp.sid(+)
and px.serial# = pp.serial#(+)
and ssw.sid = s.sid
and ssw.inst_id = s.inst_id
order by decode(px.QCINST_ID, NULL, px.INST_ID, px.QCINST_ID),
px.QCSID,
decode(px.SERVER_GROUP, NULL, 0, px.SERVER_GROUP),
px.SERVER_SET,
px.INST_ID /
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.
2. shows for the PX Deq events the processes that are exchange data.
select sw.SID as RCVSID,
decode(pp.server_name, NULL, 'A QC', pp.server_name) as RCVR,
sw.inst_id as RCVRINST,
case sw.state
WHEN 'WAITING' THEN
substr(sw.event, 1, 30)
ELSE
NULL
end as wait_event,
decode(bitand(p1, 65535),
65535,
'QC',
'P' || to_char(bitand(p1, 65535), 'fm000')) as SNDR,
bitand(p1, 16711680) - 65535 as SNDRINST,
decode(bitand(p1, 65535),
65535,
ps.qcsid,
(select sid
from gv$px_process
where server_name =
'P' || to_char(bitand(sw.p1, 65535), 'fm000')
and inst_id = bitand(sw.p1, 16711680) - 65535)) as SNDRSID,
decode(sw.state, 'WAITING', 'WAIT', 'NOT WAIT') as STATE
from gv$session_wait sw, gv$px_process pp, gv$px_session ps
where sw.sid = pp.sid(+)
and sw.inst_id = pp.inst_id(+)
and sw.sid = ps.sid(+)
and sw.inst_id = ps.inst_id(+)
and p1text = 'sleeptime/senderid'
and bitand(p1, 268435456) = 268435456
order by decode(ps.QCINST_ID, NULL, ps.INST_ID, ps.QCINST_ID),
ps.QCSID,
decode(ps.SERVER_GROUP, NULL, 0, ps.SERVER_GROUP),
ps.SERVER_SET,
ps.INST_ID
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.
- 34.
- 35.
3. shows for long running processes what are the slaves do.
select decode(px.qcinst_id,
NULL,
username,
' - ' ||
lower(substr(pp.SERVER_NAME, length(pp.SERVER_NAME) - 4, 4))) "Username",
decode(px.qcinst_id, NULL, 'QC', '(Slave)') "QC/Slave",
to_char(px.server_set) "SlaveSet",
to_char(px.inst_id) "Slave INST",
substr(opname, 1, 30) operation_name,
substr(target, 1, 30) target,
sofar,
totalwork,
units,
start_time,
timestamp,
decode(px.qcinst_id, NULL, to_char(s.sid), px.qcsid) "QC SID",
to_char(px.qcinst_id) "QC INST"
from gv$px_session px, gv$px_process pp, gv$session_longops s
where px.sid = s.sid
and px.serial# = s.serial#
and px.inst_id = s.inst_id
and px.sid = pp.sid(+)
and px.serial# = pp.serial#(+)
order by decode(px.QCINST_ID, NULL, px.INST_ID, px.QCINST_ID),
px.QCSID,
decode(px.SERVER_GROUP, NULL, 0, px.SERVER_GROUP),
px.SERVER_SET,
px.INST_ID
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
