查看会话内存占用情况
SELECT server "连接类型",
s.username,
OSUSEr,
NAME,
VALUE/1024/1024 "占用内存MB",
s.SID "会话ID",
s.serial#,
spid "操作系统进程ID",
p.PGA_USED_MEM,
p.PGA_ALLOC_MEM,
p.PGA_FREEABLE_MEM,
p.PGA_MAX_MEM
FROM v$session s, v$sesstat st, v$statname sn, v$process p
WHERE st.SID = s.SID
AND st.statistic# = sn.statistic#
AND sn.NAME LIKE 'session pga memory'
AND p.addr = s.paddr
ORDER BY VALUE DESC
如果想杀死指定会话alter system kill session 'sid,serial#';,内存没有释放执行如下步骤;
select a.spid,b.sid,b.serial#,b.username from v$process a,v$session b where a.addr=b.paddr and b.status='KILLED' ;
查看killed状态的会话,
windows下:orakill.exe orcl spid,其中orcl是数据库的sid,spid是上面查询出来的
linux下:kill -9 spid;
SQL消耗资源情况查询:
从V$SQLAREA中查询最占用资源的查询
select b.username username,a.disk_reads reads,
a.executions exec,a.disk_reads/decode(a.executions,0,1,a.executions) rds_exec_ratio,
a.sql_text Statement
from v$sqlarea a,dba_users b
where a.parsing_user_id=b.user_id
and a.disk_reads > 100000
order by a.disk_reads desc;
用buffer_gets列来替换disk_reads列可以得到占用最多内存的sql语句的相关信息。
V$SQL是内存共享SQL区域中已经解析的SQL语句。
列出使用频率最高的5个查询
select sql_text,executions
from (select sql_text,executions,
rank() over
(order by executions desc) exec_rank
from v$sql)
where exec_rank <=5;
消耗磁盘读取最多的sql top5
select disk_reads,sql_text
from (select sql_text,disk_reads,
dense_rank() over
(order by disk_reads desc) disk_reads_rank
from v$sql)
where disk_reads_rank <=5;
找出需要大量缓冲读取(逻辑读)操作的查询
select buffer_gets,sql_text
from (select sql_text,buffer_gets,
dense_rank() over
(order by buffer_gets desc) buffer_gets_rank
from v$sql)
where buffer_gets_rank<=5;
时间占用多的:
SELECT S.SQL_TEXT,
S.SQL_FULLTEXT,
S.SQL_ID,
ROUND(ELAPSED_TIME / 1000000 / (CASE
WHEN (EXECUTIONS = 0 OR NVL(EXECUTIONS, 1) = 1) THEN
1
ELSE
EXECUTIONS
END),
2) "执行时间'S'",
S.EXECUTIONS "执行次数",
S.OPTIMIZER_COST "COST",
S.SORTS,
S.MODULE, --连接模式(JDBC THIN CLIENT:程序)
-- S.LOCKED_TOTAL,
--S.PHYSICAL_READ_BYTES "物理读",
-- S.PHYSICAL_READ_REQUESTS "物理读请求",
--S.PHYSICAL_WRITE_REQUESTS "物理写",
-- S.PHYSICAL_WRITE_BYTES "物理写请求",
S.ROWS_PROCESSED "返回行数",
S.DISK_READS "磁盘读",
S.DIRECT_WRITES "直接路径写",
S.PARSING_SCHEMA_NAME,
S.LAST_ACTIVE_TIME
FROM GV$SQLAREA S
WHERE ROUND(ELAPSED_TIME / 1000000 / (CASE
WHEN (EXECUTIONS = 0 OR NVL(EXECUTIONS, 1) = 1) THEN
1
ELSE
EXECUTIONS
END),
2) > 5 --100 0000微秒=1S
AND S.PARSING_SCHEMA_NAME = USER
AND TO_CHAR(S.LAST_LOAD_TIME, 'YYYY-MM-DD') =
TO_CHAR(SYSDATE, 'YYYY-MM-DD')
AND S.COMMAND_TYPE IN (2, 3, 5, 6, 189)
ORDER BY "执行时间'S'" DESC;