Shulou(Shulou.com)06/01 Report--

Since 11G, the installation of RAC has become a physical task, but after the installation of RAC, how to ensure the stable operation of the system and how to get the performance of the system has a great impact on the stable operation of the online system in the later stage.

The following is a summary of some experience in the implementation of the project in the past year or so.

#

# # 0. Environmental inspection

# # 1. IO performance Test

# # 2. Network performance testing

# # 3. High availability test

# # 4. HP-UX hardware View Command

# # 5. Log collection

# # Note: this document is based on the HP-UX platform. Please refer to the corresponding operation commands for different platforms.

#

=

# # 0. Environmental inspection

=

# 0.1 system time / time Zone

# date

# cat / etc/TIMEZONE

# cat / etc/default/tz

# echo $TZ

# 0.2 detect kernel parameters / asynchronous IO/ network parameters

# correct value of asynchronous IO file permissions: oracle/oinstall 660101 0x000104 / dev/async

-- HP

# kctune | sort

# ls-la / dev/async

# cat / etc/rc.config.d/nddconf

# netstat-inw

-- Linux

# sysctl-a | sort

# 0.3 Trust check

$more / etc/hosts | grep-Ev'^ # | ^ $| 127.0.0.1 | vip | scan |:'| awk'{print "ssh" $2 "date;"}'> ping.sh

$sh. / ping.sh

# 0.4 Cluster status check

# crsctl stat res-t

# olsnodes-n-I-s-t

# olsnodes-c

# oifcfg getif

# 0.5 check the running status of the instance

SQL >

Set linesize 160

Select inst_id,instance_name, to_char (startup_time, 'yyyy-mm-dd hh34:mi:ss') START_TIME, status from gv$instance

# 0.6 check ASM DG capacity

SQL >

Select group_number

Name

State

Type

ROUND (total_mb / 1024, 2) TOTAL_GB

ROUND (free_mb / 1024, 2) FREE_GB

Round ((total_mb-free_mb) / (case total_mb)

When 0 then

one

Else

Total_mb

End) * 100

2) ratio

From v$asm_diskgroup

=

# # 1. IO performance Test

=

# 1.1 calibrate_io Test

SELECT D.NAME

I.ASYNCH_IO

FROM V$DATAFILE D

V$IOSTAT_FILE I

WHERE D.FILE# = I.FILE_NO

AND I.FILETYPE_NAME = 'Data File'

-- Enabled both direct I and asynchronous O where possible.

ALTER SYSTEM SET filesystemio_options=setall SCOPE=SPFILE

SET SERVEROUTPUT ON

DECLARE

L_latency PLS_INTEGER

L_iops PLS_INTEGER

L_mbps PLS_INTEGER

BEGIN

DBMS_RESOURCE_MANAGER.CALIBRATE_IO (num_physical_disks = > 26,-- > number of disks select count (1) from v$asm_disk

Max_latency = > 20,-- > maximum delay

Max_iops = > l_iops

Max_mbps = > l_mbps

Actual_latency = > l_latency)

DBMS_OUTPUT.put_line ('Max IOPS =' | | l_iops)

DBMS_OUTPUT.put_line ('Max MBPS =' | | l_mbps)

DBMS_OUTPUT.put_line ('Latency =' | | l_latency)

END

/

-- Monitoring the rate of IO and HBA cards

# sar-H 2 10000

# 1.2 fio testing

=

# # 2. Network performance testing

=

# 2.1 netperf Test

Cd / usr/contrib/bin/tools

-- server side

# netserver-4-L 192.168.30.76-p 8000

-- client

# netperf-H 192.168.30.76-p 8000-n 20-l 60

-- Test UDP delivery

# netperf-t UDP_RR-H 192.168.30.76-p 8000-l 10-r 1500 1500

# netperf-t UDP_RR-H 192.168.30.76-p 8000-l 10-r 9000 9000

# # 2.2 iperf test, iperf3 in Linux

-- server side

# iperf-s-p 5001-I 2

-- client

# iperf-c 192.168.30.76-P 4-t 30-I 2-p 5001

# # 2.3 ftp rate Test

Ftp > put "| dd if=/dev/zero bs=32k count=10000" / dev/null

# 2.4 MUT test is 9000 (root permission is required for tcpdump)

-- lan902 heartbeat network card

-- 192.168.20.67 Peer heartbeat Nic address

-- establish a tcp connection from this node by accessing each other's heartbeat IP, and then you can see the MSs value by looking at the TCP packet captured above. If it is 8960, then the mtu is 9000.

-- send a package test through netperf or ping. The ICMP length is 8976.

Cd / usr/contrib/bin/tools

# tcpdump-I lan902 ip host 192.168.20.67

# telnet 192.168.20.67 22

# netperf-H 192.168.20.67-p 8000-n 20-l 60

# ping 192.168.20.67 9000

=

# # 3. High availability test

=

# 3.1 Storage Link Test

# View the link disabled and endable status in the operating system log, and view the data traffic information of HBA card

# View disk link aggregation

# during the storage link test, turn on DBMS_RESOURCE_MANAGER.CALIBRATE_IO for pressurization

-- confirm the device link

# ioscan-m dsf / dev/rdisk/disk101

# tail-f / var/adm/syslog/syslog.log

# sar-H 2 10000

# 3.2 Network Card / switch Test

# check the cluster log for network outages, and view the ASM alarm log (grid users)

# check the aggregation of the network card, check the log of the network card, and see if packets are lost

# during network testing, turn on netperf to send packets for pressurization

Grid$ cd $ORACLE_HOME/log/ `hostname` /; tail-f alert`hostname`.log

Grid$ cd $ORACLE_BASE/diag/asm/+asm/$ORACLE_SID/trace; tail-f alert_$ORACLE_SID.log

# glance-> l

# nwmgr-S apa-c lan902-v

# cd / var/adm; netfmt-f nettl.LOG000

# ping IP

=

# # 4. HP-UX hardware View Command

=

# # 4.1 View Devices

-- FC equipment

# ioscan-fnCfc

-- Network card equipment

# ioscan-fnClan

# # 4.2 View FC device status

# fcmsutil / dev/fcd1

# # 4.3 View disk multipath information

# ioscan-m dsf / dev/rdisk/disk101

# # 4.4 View Nic information

# lanscan-qv

# netstat-inw

# # 4.5 View the binding information of the Nic

-- Network card binding information

# nwmgr-S apa-c lan902-v

-- Network card priority

# nwmgr-g-An all-c lan5-S apa

=

# # 5. Log collection

=

# # 5.1 ASM Log (grid)

Cd $ORACLE_BASE/diag/asm/+asm/$ORACLE_SID/trace; tail-f alert_$ORACLE_SID.log

# # 5.2 Cluster logs (grid)

Cd $ORACLE_HOME/log/ `hostname` /; tail-f alert`hostname`.log

# # 5.3 Database Log (oracle)

Cd $ORACLE_BASE/diag/rdbms/ `echo ${ORACLE_SID%?} `/ $ORACLE_SID/trace; tail-f alert_$ORACLE_SID.log

# # 5.4 operating system log (root)

Tail-f / var/adm/syslog/syslog.log

# # 5.5 Network Card Log (root)

Cd / var/adm; netfmt-f nettl.LOG000

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.