【Oracle】 检查log fie sysnc 等待事件的脚本

本文涉及的产品
日志服务 SLS,月写入数据量 50GB 1个月
简介: -- NAME: LFSDIAG.SQL -- ------------------------------------------------------------------------ -- AUTHOR: Michael Polaski - Ora...
-- NAME: LFSDIAG.SQL 
-- ------------------------------------------------------------------------ 
-- AUTHOR: Michael Polaski - Oracle Support Services 
-- ------------------------------------------------------------------------ 
-- PURPOSE: 
-- This script. is intended to provide a user friendly guide to troubleshoot 
-- log file sync waits. The script. will look at important parameters involved 
-- in log file sync waits, log file sync wait histogram data, and the script 
-- will look at the worst average log file sync times in the active session 
-- history data and AWR data and dump information to help determine why those 
-- times were the highest. The script. will create a file called 
-- lfsdiag_.out in your local directory. 

set echo off 
set feedback off 
column timecol new_value timestamp 
column spool_extension new_value suffix 
select to_char(sysdate, 'Mondd_hh24mi') timecol, '.out' spool_extension
  from sys.dual;
column output new_value dbname
  select value || '_' output from v$parameter where name = 'db_name';
spool lfsdiag_&&dbname&&timestamp&&suffix 
set trim on 
set trims on 
set lines 130 
set pages 100 
set verify off 
alter session set optimizer_features_enable = '10.2.0.4'; 

PROMPT LFSDIAG DATA FOR &&dbname&&timestamp 
PROMPT Note: All timings are in milliseconds (1000 milliseconds = 1 second) 

PROMPT 
PROMPT IMPORTANT PARAMETERS RELATING TO LOG FILE SYNC WAITS: 
column name format a40 wra 
column value format a40 wra 
select inst_id, name, value
  from gv$parameter
 where ((value is not null and name like '%log_archive%') or
       name like '%commit%' or name like '%event=%' or name like '%lgwr%')
   and name not in
       (select name
          from gv$parameter
         where (name like '%log_archive_dest_state%' and value = 'enable')
            or name = 'log_archive_format')
 order by 1, 2, 3;

PROMPT 
PROMPT HISTOGRAM DATA FOR LFS AND OTHER RELATED WAITS: 
PROMPT 
PROMPT APPROACH: Look at the wait distribution for log file sync waits 
PROMPT by looking at "wait_time_milli". Look at the high wait times then 
PROMPT see if you can correlate those with other related wait events. 
column event format a40 wra 
select inst_id, event, wait_time_milli, wait_count
  from gv$event_histogram
 where event in ('log file sync',
                 'gcs log flush sync',
                 'log file parallel write',
                 'wait for scn ack',
                 'log file switch completion',
                 'gc cr grant 2-way',
                 'gc buffer busy',
                 'gc current block 2-way')
    or event like '%LGWR%'
    or event like '%LNS%'
 order by 2 desc, 1, 3;

PROMPT 
PROMPT ORDERED BY WAIT_TIME_MILLI 
select inst_id, event, wait_time_milli, wait_count
  from gv$event_histogram
 where event in ('log file sync',
                 'gcs log flush sync',
                 'log file parallel write',
                 'wait for scn ack',
                 'log file switch completion',
                 'gc cr grant 2-way',
                 'gc buffer busy',
                 'gc current block 2-way')
    or event like '%LGWR%'
    or event like '%LNS%'
 order by 3, 1, 2 desc;

PROMPT 
PROMPT REDO WRITE STATS 
PROMPT 
PROMPT "redo write time" in centiseconds (100 per second) 
PROMPT 11.1: "redo write broadcast ack time" in centiseconds (100 per second) 
PROMPT 11.2: "redo write broadcast ack time" in microseconds (1000 per millisecond) 
column value format 99999999999999999999 
column milliseconds format 99999999999999.999 
select v.version,
       ss.inst_id,
       ss.name,
       ss.value,
       decode(substr(version, 1, 4),
              '11.1',
              decode(name,
                     'redo write time',
                     value * 10,
                     'redo write broadcast ack time',
                     value * 10),
              '11.2',
              decode(name,
                     'redo write time',
                     value * 10,
                     'redo write broadcast ack time',
                     value / 1000),
              decode(name, 'redo write time', value * 10)) milliseconds
  from gv$sysstat ss, v$instance v
 where name like 'redo write%'
   and value > 0
 order by 1, 2, 3;

PROMPT 
PROMPT ASH THRESHOLD... 
PROMPT 
PROMPT This will be the threshold in milliseconds for average log file sync 
PROMPT times. This will be used for the next queries to look for the worst 
PROMPT 'log file sync' minutes. Any minutes that have an average log file 
PROMPT sync time greater than the threshold will be analyzed further. 
column threshold_in_ms new_value threshold format 999999999.999 
select min(threshold_in_ms) threshold_in_ms
  from (select inst_id,
               to_char(sample_time, 'Mondd_hh24mi') minute,
               avg(time_waited) / 1000 threshold_in_ms
          from gv$active_session_history
         where event = 'log file sync'
         group by inst_id, to_char(sample_time, 'Mondd_hh24mi')
         order by 3 desc)
 where rownum

PROMPT 
PROMPT ASH WORST MINUTES FOR LOG FILE SYNC WAITS: 
PROMPT 
PROMPT APPROACH: These are the minutes where the avg log file sync time 
PROMPT was the highest (in milliseconds). 
column event format a30 tru 
column program format a35 tru 
column total_wait_time format 999999999999.999 
column avg_time_waited format 999999999999.999 
select to_char(sample_time, 'Mondd_hh24mi') minute,
       inst_id,
       event,
       sum(time_waited) / 1000 TOTAL_WAIT_TIME,
       count(*) WAITS,
       avg(time_waited) / 1000 AVG_TIME_WAITED
  from gv$active_session_history
 where event = 'log file sync'
 group by to_char(sample_time, 'Mondd_hh24mi'), inst_id, event
having avg(time_waited) / 1000 > &&threshold
 order by 1, 2;

PROMPT 
PROMPT ASH LFS BACKGROUND PROCESS WAITS DURING WORST MINUTES: 
PROMPT 
PROMPT APPROACH: What is LGWR doing when 'log file sync' waits 
PROMPT are happening? LMS info may be relevent for broadcast 
PROMPT on commit and LNS data may be relevant for dataguard. 
PROMPT If more details are needed see the ASH DETAILS FOR WORST 
PROMPT MINUTES section at the bottom of the report. 
column inst format 999 
column event format a30 tru 
column program format a35 wra 
select to_char(sample_time, 'Mondd_hh24mi') minute,
       inst_id inst,
       program,
       event,
       sum(time_waited) / 1000 TOTAL_WAIT_TIME,
       count(*) WAITS,
       avg(time_waited) / 1000 AVG_TIME_WAITED
  from gv$active_session_history
 where to_char(sample_time, 'Mondd_hh24mi') in
       (select to_char(sample_time, 'Mondd_hh24mi')
          from gv$active_session_history
         where event = 'log file sync'
         group by to_char(sample_time, 'Mondd_hh24mi'), inst_id
        having avg(time_waited) / 1000 > &&threshold)
   and (program like '%LGWR%' or program like '%LMS%' or
        program like '%LNS%' or event = 'log file sync')
 group by to_char(sample_time, 'Mondd_hh24mi'), inst_id, program, event
 order by 1, 2, 3, 5 desc, 4;

PROMPT 
PROMPT AWR WORST AVG LOG FILE SYNC SNAPS: 
PROMPT 
PROMPT APPROACH: These are the AWR snaps where the average 'log file sync' 
PROMPT times were the highest. 
column begin format a12 tru 
column end format a12 tru 
column name format a13 tru 
select 
  dhs.snap_id, 
  dhs.instance_number inst, 
  to_char(dhs.begin_interval_time,'Mondd_hh24mi') BEGIN, 
  to_char(dhs.end_interval_time,'Mondd_hh24mi') END, 
  en.name, 
  se.time_waited_micro/1000 total_wait_time, 
  se.total_waits, 
  se.time_waited_micro/1000 / se.total_waits avg_time_waited 
from 
  dba_hist_snapshot dhs, 
  wrh$_system_event se, 
  v$event_name en 
where (dhs.snap_id = se.snap_id and dhs.instance_number = se.instance_number) 
  and se.event_id = en.event_id and en.name = 'log file sync' and 
  dhs.snap_id in (
select snap_id
  from (select se.snap_id,
               se.time_waited_micro / 1000 / se.total_waits avg_time_waited
          from wrh$_system_event se, v$event_name en
         where se.event_id = en.event_id
           and en.name = 'log file sync'
         order by avg_time_waited desc)
 where rownum
order by 1,2; 

PROMPT 
PROMPT AWR REDO WRITE STATS 
PROMPT 
PROMPT "redo write time" in centiseconds (100 per second) 
PROMPT 11.1: "redo write broadcast ack time" in centiseconds (100 per second) 
PROMPT 11.2: "redo write broadcast ack time" in microseconds (1000 per millisecond) 
column stat_name format a30 tru 
select v.version,
       ss.snap_id,
       ss.instance_number inst,
       sn.stat_name,
       ss.value,
       decode(substr(version, 1, 4),
              '11.1',
              decode(stat_name,
                     'redo write time',
                     value * 10,
                     'redo write broadcast ack time',
                     value * 10),
              '11.2',
              decode(stat_name,
                     'redo write time',
                     value * 10,
                     'redo write broadcast ack time',
                     value / 1000),
              decode(stat_name, 'redo write time', value * 10)) milliseconds
  from wrh$_sysstat ss, wrh$_stat_name sn, v$instance v
 where ss.stat_id = sn.stat_id
   and sn.stat_name like 'redo write%'
   and ss.value > 0
   and ss.snap_id in
       (select snap_id
          from (select se.snap_id,
                       se.time_waited_micro / 1000 / se.total_waits avg_time_waited
                  from wrh$_system_event se, v$event_name en
                 where se.event_id = en.event_id
                   and en.name = 'log file sync'
                 order by avg_time_waited desc)
         where rownum
 order by 1, 2, 3;

PROMPT 
PROMPT AWR LFS AND OTHER RELATED WAITS FOR WORST LFS AWRs: 
PROMPT 
PROMPT APPROACH: These are the AWR snaps where the average 'log file sync' 
PROMPT times were the highest. Look at related waits at those times. 
column name format a40 tru 
select se.snap_id,
       se.instance_number inst,
       en.name,
       se.total_waits,
       se.time_waited_micro / 1000 total_wait_time,
       se.time_waited_micro / 1000 / se.total_waits avg_time_waited
  from wrh$_system_event se, v$event_name en
 where se.event_id = en.event_id
   and (en.name in ('log file sync',
                    'gcs log flush sync',
                    'log file parallel write',
                    'wait for scn ack',
                    'log file switch completion',
                    'gc cr grant 2-way',
                    'gc buffer busy',
                    'gc current block 2-way') or en.name like '%LGWR%' or
       en.name like '%LNS%')
   and se.snap_id in
       (select snap_id
          from (select se.snap_id,
                       se.time_waited_micro / 1000 / se.total_waits avg_time_waited
                  from wrh$_system_event se, v$event_name en
                 where se.event_id = en.event_id
                   and en.name = 'log file sync'
                 order by avg_time_waited desc)
         where rownum
 order by 1, 6 desc;

PROMPT 
PROMPT AWR HISTOGRAM DATA FOR LFS AND OTHER RELATED WAITS FOR WORST LFS AWRs: 
PROMPT Note: This query won't work on 10.2 - ORA-942 
PROMPT 
PROMPT APPROACH: Look at the wait distribution for log file sync waits 
PROMPT by looking at "wait_time_milli". Look at the high wait times then 
PROMPT see if you can correlate those with other related wait events. 
select eh.snap_id,
       eh.instance_number inst,
       en.name,
       eh.wait_time_milli,
       eh.wait_count
  from wrh$_event_histogram eh, v$event_name en
 where eh.event_id = en.event_id
   and (en.name in ('log file sync',
                    'gcs log flush sync',
                    'log file parallel write',
                    'wait for scn ack',
                    'log file switch completion',
                    'gc cr grant 2-way',
                    'gc buffer busy',
                    'gc current block 2-way') or en.name like '%LGWR%' or
       en.name like '%LNS%')
   and snap_id in
       (select snap_id
          from (select se.snap_id,
                       se.time_waited_micro / 1000 / se.total_waits avg_time_waited
                  from wrh$_system_event se, v$event_name en
                 where se.event_id = en.event_id
                   and en.name = 'log file sync'
                 order by avg_time_waited desc)
         where rownum
 order by 1, 3 desc, 2, 4;

PROMPT 
PROMPT ORDERED BY WAIT_TIME_MILLI 
PROMPT Note: This query won't work on 10.2 - ORA-942 
select eh.snap_id,
       eh.instance_number inst,
       en.name,
       eh.wait_time_milli,
       eh.wait_count
  from wrh$_event_histogram eh, v$event_name en
 where eh.event_id = en.event_id
   and (en.name in ('log file sync',
                    'gcs log flush sync',
                    'log file parallel write',
                    'wait for scn ack',
                    'log file switch completion',
                    'gc cr grant 2-way',
                    'gc buffer busy',
                    'gc current block 2-way') or en.name like '%LGWR%' or
       en.name like '%LNS%')
   and snap_id in
       (select snap_id
          from (select se.snap_id,
                       se.time_waited_micro / 1000 / se.total_waits avg_time_waited
                  from wrh$_system_event se, v$event_name en
                 where se.event_id = en.event_id
                   and en.name = 'log file sync'
                 order by avg_time_waited desc)
         where rownum
 order by 1, 4, 2, 3 desc;

PROMPT 
PROMPT ASH DETAILS FOR WORST MINUTES: 
PROMPT 
PROMPT APPROACH: If you cannot determine the problem from the data 
PROMPT above, you may need to look at the details of what each session 
PROMPT is doing during each 'bad' snap. Most likely you will want to 
PROMPT note the times of the high log file sync waits, look at what 
PROMPT LGWR is doing at those times, and go from there... 
column program format a45 wra 
column sample_time format a25 tru 
column event format a30 tru 
column time_waited format 999999.999 
column p1 format a40 tru 
column p2 format a40 tru 
column p3 format a40 tru 
select sample_time,
       inst_id inst,
       session_id,
       program,
       event,
       time_waited / 1000 TIME_WAITED,
       p1text || ': ' || p1 p1,
       p2text || ': ' || p2 p2,
       p3text || ': ' || p3 p3
  from gv$active_session_history
 where to_char(sample_time, 'Mondd_hh24mi') in
       (select to_char(sample_time, 'Mondd_hh24mi')
          from gv$active_session_history
         where event = 'log file sync'
         group by to_char(sample_time, 'Mondd_hh24mi'), inst_id
        having avg(time_waited) / 1000 > &&threshold)
 order by 1, 2, 3, 4, 5;

select to_char(sysdate,'Mondd hh24:mi:ss') TIME from dual; 

spool off 

PROMPT 
PROMPT OUTPUT FILE IS: lfsdiag_&&dbname&&timestamp&&suffix 
PROMPT 
阅读(5359) | 评论(0) | 转发(0) |
给主人留下些什么吧!~~
评论热议
请登录后评论。

登录 注册

相关实践学习
日志服务之使用Nginx模式采集日志
本文介绍如何通过日志服务控制台创建Nginx模式的Logtail配置快速采集Nginx日志并进行多维度分析。
目录
相关文章
|
Oracle 前端开发 关系型数据库
在Oracle的ADR中设置自动删除trace文件的策略
姚远在一个有两万个客户的公司做数据库支持,什么稀奇古怪的事情都能遇到,有个客户的数据库不停地产生大量的trace,经常把硬盘撑爆,看看姚远怎么解决这个问题的。
116 0
|
Oracle 关系型数据库
Oracle trace文件的清理
Oracle trace文件的清理
584 0
Oracle trace文件的清理
|
Oracle 关系型数据库
Oracle修改redo log大小的方法
目的:修改当前在线日志从默认50M增加至512M。 1.查看当前日志组的状态 SQL> select group#,members,bytes/1024/1024,status from v$log;     GROUP#    MEMBERS BYTES/1024/1024 ...
1984 0
|
Oracle 关系型数据库 Shell