删除11.2.0.4 rac中 ora.crf的两类文件 和 禁止ora.crf随ohas启动而启动

删除11.2.0.4 rac中 ora.crf的两类文件 和  禁止ora.crf随ohas启动而启动

背景：
ora.crf服务是为Cluster Health Monitor（以下简称CHM）提供服务的，用来自动收集操作系统的资源（CPU、内存、SWAP、进程、I/O以及网络等）的使用情况。由于bug 10165314，ora.crf服务生成的文件（$GI_HOME/crf/db/$HOMENAME/目录中所有的crf*.bdb和$HOSTNAME.ldb）会很大，这样就对$GI_HOME的使用率造成了压力。

下面来说明删除删除这两类文件（$GI_HOME/crf/db/$HOMENAME/目录中所有的crf*.bdb和$HOSTNAME.ldb）的步骤 和 禁止ora.crf随ohas启动而启动的步骤。

一、删除crf*.bdb和$HOSTNAME.ldb的步骤

[root@leihost1 leihost1]# /u01/11.2.0/grid/bin/crsctl stop res ora.crf -init
CRS-2673: Attempting to stop 'ora.crf' on 'leihost1'
CRS-2677: Stop of 'ora.crf' on 'leihost1' succeeded
[root@leihost1 leihost1]# 
[root@leihost1 leihost1]# cd /u01/11.2.0/grid/crf/db/leihost1
[root@leihost1 leihost1]# ls -lrt
total 195748
-rw-r--r-- 1 root root 120000000 Oct 24 20:20 leihost1.ldb
-rw-r--r-- 1 root root   1063396 Oct 24 20:27 24-OCT-2015-20:27:46.txt
-rw-r----- 1 root root      8192 Oct 24 20:28 crfconn.bdb
-rw-r----- 1 root root  16777216 Oct 25 11:06 log.0000000018
-rw-r----- 1 root root     24576 Oct 25 11:37 __db.001
-rw-r----- 1 root root      8192 Oct 25 11:38 repdhosts.bdb
-rw-r----- 1 root root  16777216 Oct 25 11:38 log.0000000019
-rw-r----- 1 root root   2301952 Oct 25 11:38 crfts.bdb
-rw-r----- 1 root root   3522560 Oct 25 11:38 crfloclts.bdb
-rw-r----- 1 root root   2621440 Oct 25 11:38 crfhosts.bdb
-rw-r----- 1 root root   3686400 Oct 25 11:38 crfcpu.bdb
-rw-r----- 1 root root 143720448 Oct 25 11:38 crfclust.bdb
-rw-r----- 1 root root   3244032 Oct 25 11:38 crfalert.bdb
-rw-r----- 1 root root     57344 Oct 25 11:38 __db.006
-rw-r----- 1 root root    401408 Oct 25 11:38 __db.002
-rw-r----- 1 root root   1187840 Oct 25 11:38 __db.005
-rw-r----- 1 root root   2162688 Oct 25 11:38 __db.004
-rw-r----- 1 root root   2629632 Oct 25 11:38 __db.003
[root@leihost1 leihost1]# rm *.bdb
rm: remove regular file `crfalert.bdb'? y
rm: remove regular file `crfclust.bdb'? y
rm: remove regular file `crfconn.bdb'? y
rm: remove regular file `crfcpu.bdb'? y
rm: remove regular file `crfhosts.bdb'? y
rm: remove regular file `crfloclts.bdb'? y
rm: remove regular file `crfts.bdb'? y
rm: remove regular file `repdhosts.bdb'? y
[root@leihost1 leihost1]# du -sh
40M .
[root@leihost1 leihost1]# ls -lrt
total 40200
-rw-r--r-- 1 root root 120000000 Oct 24 20:20 leihost1.ldb
-rw-r--r-- 1 root root   1063396 Oct 24 20:27 24-OCT-2015-20:27:46.txt
-rw-r----- 1 root root  16777216 Oct 25 11:06 log.0000000018
-rw-r----- 1 root root     24576 Oct 25 11:37 __db.001
-rw-r----- 1 root root  16777216 Oct 25 11:38 log.0000000019
-rw-r----- 1 root root     57344 Oct 25 11:38 __db.006
-rw-r----- 1 root root    401408 Oct 25 11:38 __db.002
-rw-r----- 1 root root   1187840 Oct 25 11:38 __db.005
-rw-r----- 1 root root   2162688 Oct 25 11:38 __db.004
-rw-r----- 1 root root   2629632 Oct 25 11:38 __db.003
[root@leihost1 leihost1]# mv leihost1.ldb back_leihost1.ldb
[root@leihost1 leihost1]# ls -lrt
total 40200
-rw-r--r-- 1 root root 120000000 Oct 24 20:20 back_leihost1.ldb
-rw-r--r-- 1 root root   1063396 Oct 24 20:27 24-OCT-2015-20:27:46.txt
-rw-r----- 1 root root  16777216 Oct 25 11:06 log.0000000018
-rw-r----- 1 root root     24576 Oct 25 11:37 __db.001
-rw-r----- 1 root root  16777216 Oct 25 11:38 log.0000000019
-rw-r----- 1 root root     57344 Oct 25 11:38 __db.006
-rw-r----- 1 root root    401408 Oct 25 11:38 __db.002
-rw-r----- 1 root root   1187840 Oct 25 11:38 __db.005
-rw-r----- 1 root root   2162688 Oct 25 11:38 __db.004
-rw-r----- 1 root root   2629632 Oct 25 11:38 __db.003
[root@leihost1 leihost1]# /u01/11.2.0/grid/bin/crsctl status res ora.crf -init
NAME=ora.crf
TYPE=ora.crf.type
TARGET=OFFLINE
STATE=OFFLINE
[root@leihost1 leihost1]#

如上摘自：

ODA Nodes Lacking Space Due to Large Cluster Health Monitor FileCrfclust.Bdb (文档 ID 1616910.1)

 

如上步骤可以在clusterware和database 运行的时候操作，原因是：

Is stop/start ora.crf affecting clusterware function or cluster database function?

No, stop/start ora.crf resource will stop and start Cluster Health Monitor and its data collection, it will not affect clusterware or database functionality.

如上摘自：
Cluster Health Monitor (CHM) FAQ (文档 ID 1328466.1)

上面的步骤，请在集群的其他节点上也执行。

 

二、
禁止ora.crf随ohas启动而启动的步骤。
2.1

[root@leihost1 leihost1]# /u01/11.2.0/grid/bin/crsctl modify res ora.crf -attr ENABLED=0 -init

2.2查看修改后的效果：

[root@leihost1 leihost1]# /u01/11.2.0/grid/bin/crsctl  status res ora.crf -init -f
NAME=ora.crf
TYPE=ora.crf.type
STATE=OFFLINE
TARGET=ONLINE
ACL=owner:root:rw-,pgrp:oinstall:rw-,other::r--,user:grid:r-x
ACTION_FAILURE_TEMPLATE=
ACTION_SCRIPT=
ACTIVE_PLACEMENT=0
AGENT_FILENAME=%CRS_HOME%/bin/orarootagent%CRS_EXE_SUFFIX%
AUTO_START=always
CARDINALITY=1
CARDINALITY_ID=0
CHECK_ARGS=
CHECK_COMMAND=
CHECK_INTERVAL=30
CLEAN_ARGS=
CLEAN_COMMAND=
CREATION_SEED=16
DAEMON_LOGGING_LEVELS=CRFMOND=0,CRFLDREP=0,CRFLOGD=0,CRFPROXY=0,OCLUMON=0,OCRAPI=0,OCRCLI=0,OCRMSG=0,CSSCLNT=0,CRFM=0
DAEMON_TRACING_LEVELS=CRFMOND=0,CRFLDREP=0,CRFLOGD=0,CRFPROXY=0,OCLUMON=0,OCRAPI=0,OCRCLI=0,OCRMSG=0,CSSCLNT=0,CRFM=0
DEFAULT_TEMPLATE=
DEGREE=1
DESCRIPTION="Resource type for Crf Agents"
DETACHED=true
ENABLED=0                 --------------->此处
FAILOVER_DELAY=0
FAILURE_INTERVAL=3
FAILURE_THRESHOLD=5
HOSTING_MEMBERS=
ID=ora.crf
LOAD=1
LOGGING_LEVEL=1
NOT_RESTARTING_TEMPLATE=
OFFLINE_CHECK_INTERVAL=0
ORA_VERSION=11.2.0.4.0
PID_FILE=
PLACEMENT=balanced
PROCESS_TO_MONITOR=
PROFILE_CHANGE_TEMPLATE=
RESTART_ATTEMPTS=5
SCRIPT_TIMEOUT=60
SERVER_POOLS=
START_ARGS=
START_COMMAND=
START_DEPENDENCIES=hard(ora.gpnpd)
START_TIMEOUT=120
STATE_CHANGE_TEMPLATE=
STOP_ARGS=
STOP_COMMAND=
STOP_DEPENDENCIES=hard(shutdown:ora.gipcd)
STOP_TIMEOUT=120
UNRESPONSIVE_TIMEOUT=180
UPTIME_THRESHOLD=1m
USR_ORA_ENV=

[root@leihost1 leihost1]# 

2.3 重启crs

# $GRID_ORACLE_HOME/bin/crsctl stop crs
# $GRID_ORACLE_HOME/bin/crsctl start crs

2.4 观察ora.crf的输出：

[root@leihost1 leihost1]# /u01/11.2.0/grid/bin/crsctl  status res ora.crf -init 
NAME=ora.crf
TYPE=ora.crf.type
TARGET=ONLINE
STATE=OFFLINE --->OFFLINE

2.5 观察$GI_HOME/crf/db/$HOMENAME/目录中所有的crf*.bdb和$HOSTNAME.ldb, 不再生成。

上面的步骤，请在集群的其他节点上也执行。

 

参考自：mos文章：
How to Move/Recreate GI Management Repository to Different Shared Storage (Diskgroup, CFS or NFS etc) (文档 ID 1589394.1)

该文章有这么一句话：
1. Stop and disable ora.crf resource.

On each node, as root user:
# <GI_HOME>/bin/crsctl stop res ora.crf -init
# <GI_HOME>/bin/crsctl modify res ora.crf -attr ENABLED=0 -init