CN105760262A - Method for carrying out cross validation on stability of hard disk based on dc and reboot under linux - Google Patents

Abstract

The invention provides a method for carrying out cross validation on stability of a hard disk based on dc and reboot under linux, and relates to the technical field of computer servers. A specific realizing process of the method comprises the following step: 1, at least establishing two node linux testing environments; 2, using each node not only as a server side, but also as a client side, and changing the role of each node in turns during testing; 3, carrying out dc and reboot testing on nodes under a dd pressure state; 4, detecting a slot position, the capacity and the velocity of the hard disk after completing dc and reboot under dd pressure every time, and confirming whether the working state of the hard disk is normal or not. The support of resources such as additional control nodes and interchangers are not needed, the allocation of a testing resource is optimized, and testing contents are more enriched in minimized resource allocation.

Description

A kind of based on the method for dc and reboot cross validation hard disk stability under linux

Technical field

The present invention relates to computer server technical field, particularly relate to a kind of optimum resource distribution, hard disk carries out the hard disk Detection of Stability method of dc and reboot cross-beta under dd pressure.

Background technology

In the client of current each big manufacturer server and each big IT field, the stability of hard disk is the most important thing that everybody pays close attention to all the time, and the stable work of hard disk is also the most basic guarantee of the safety to customer data.

Conventional dc test needs one control node and be with outer transmission switching on and shutting down order to realize test machine by switch, test resource is had certain requirement, will be seemed trouble when only to 2 nodes or little node test time, in addition it is also necessary to extra coordination one controls node and a switch.Reboot test is then the reboot test that continuous print is under OS, and this continuous print reboot test is single, less meets practical application scene.

Summary of the invention

In order to solve above technical problem, this paper presents a kind of based on the method for dc and reboot cross validation hard disk stability under linux.The technical assignment of the present invention is that the one for traditional test methods is warm with improvement, and simulation is to the hard disk stability verification method of Client application scene further.The present invention 2 nodes of minimum needs can be carried out dc, reboot and dd stress test.Do not need extra additional resource demand, closing to reality application scenarios after test event mixing.

The technical scheme is that

A kind of based on the method for dc and reboot cross validation hard disk stability under linux:

One, 2 node linux test environments are built: according to the actual requirements, it is possible to select different types of hard disk, SATA, SAS or SSD；

Two, configuration systematic parameter: 2 test machine are installed operating system, it is possible to install RedHat, SUSE etc., it is proposed that install completely.

Three, test environment configuration: require that 2 nodes open ipmi service, chkconfigipmion.Then row performs serviceipmistart, initializes ipmi state；

Four, hard disk Detection of Stability: after preparation completes, by dc and reboot test that the node in dd pressure is hocketed, detects the slot position of hard disk, capacity, speed and smartctllog state every time after having tested；

Need powerstatus and Total_Power is judged at dc test phase, when test node having been performed be with outer poweroff order, need to go to obtain the powerstatus of node every 20s, when powerstatus is off, then to sending poweron order outside node band；If node powerstatus is on, controls node and will carry out dd stress test, after dd pressure surveys full 120s, perform reboot.

After test node completes OSinitialization after dc, dd stress test will be entered.And judge controlling node Total_Power, if within Total_Power value ± 5% that Total_Power value is when dd pressure time, node will be performed poweroff order.If Total_Power value is less than Total_Power value-5%, by sleep10, and then obtain Total_Power value and judge.

Each slot position detecting hard disk after being inducted into OS, capacity, speed and smartctllog state.Exception record is made mark, it is simple to the inspection after test.

The described step that realizes completes in being provided with the server of operating system of linux kernel, and during operation, user carries out with root identity logs.

The invention have the advantages that

One complete cycle of one dc and reboot cross validation test, this cross-beta closing to reality Client application scene.Dc under dd pressure is restarted analog hard disk and is tackled the stability test of electricity order under extremely when high speed operation state, the abnormal restarting test in normal operation of the reboot analog hard disk under dd pressure.

Accompanying drawing explanation

Fig. 1Be the present invention realize flow processFigure。

Detailed description of the invention

Below present disclosure is carried out more detailed elaboration:

Test has been built, and system can normally detect the hard disc apparatus of node；Single-deck is detected by logical hard disk detection instrument, scan toolWhether passableProperly functioning；

1) test environment detection

Proceed to sign under OS by test machine, confirm that hard disk is current completely in place, check that hard disk state indicator lamp is all working properly；All hard disks correctly can be identified by OS, disk identifier of hard disk and hard disk SLOT one_to_one corresponding, and corresponding relation is correct；Chkconfigipmi list confirms that ipmi starts in system service.

2) test starts

According to different application scenarios, installing different operating system, the method supports the OS of RHEL, Centos, SUSE series；System is installed completely, after system installation, carries out test case as follows；

A, Node1, node2 is respectively provided with static BMCip address, writes into self-starting script at/etc/rc.local.

B, self-starting script 1 synopsis are as follows:

#/bin/bash

The current script of # is used for confirming that whether disk identifier of hard disk normal with hard disk corresponding relation after restarting every time

If [-f "/root/hdd.csv "] # judges whether that the hdd generating test records log

then

Fe=ok# normal recordings mark

else

Echo " count, sda, sdb, sdc, sdd, sde, sdf, sdg, sdh, sdi, sdj, sdk, sdl, status " > >/root/hddinfo.csv# initialize generate each drive to hdd record log

sleep1

fi

count=`cat/root/hddinfo.csv|wc-l`

foriin{a..l}

do

smartctl-i/dev/sd$i|grepSerial|awk'{print$3","}'>>/root/temp.txt

done

If [-f "/root/stander.txt "] # determines whether the correct node hard disk information record of standard

then

flag=ok

else

Cp/root/temp.txt/root/stander.txt# will be if it did not, file will be generated for standard with this

fi

The interim hard disk information that diff/root/temp.txt/root/stander.txt# produces every time and standard information compare

if[!$-eq0] # judges whether return value is 0

then

Sort=" error " # is non-zero is abnormal

else

Sort=" ok " # be 0 normal

fi

Echo " $ count, " `cat/root/temp.txt` $ sort > >/root/hddinfo.csv# is stored to each data in hddinfo.csv

Rm-rf/root/temp.txt# deletes temporary file

C, self-starting script 2 synopsis are as follows:

#!/bin/bash

Foriin{a..l}# is as the criterion with practical situation, and testing hard disk is 12 pieces herein

do

Echo " Disk/dev/sd $ i " > > root/HDD_INFO.log# print disk identifier of hard disk in log

Smartctl-i/dev/sd $ i | grep-E'SerialNumber | Capacity | Speed' > > root/cap_speed.log# obtains the string number of hard disk, capacity and speed information

done

Foriin{a..l}# is as the criterion with practical situation, and testing hard disk is 12 pieces herein

do

All of hard disk is carried out dd and reads test by ddif=/dev/sd $ iof=/dev/null&#

done

sleep30

Stand_power=`ipmitoolsdr | grepTotal_Power`# presses measurement of power to consume for standard power consumption with the dd of the machine

Stand_power_x=`echo $ stand_power*0.95 | bc`# power consumption-5%

Stand_power_y=`echo $ stand_power*1.05 | bc`# power consumption+5%

Stress_power=`ipmitool Ilanplus H192.168.1.2 Uadmin PadminTotal_Power`# is with outer initialization test node power consumption

[$ stress_power gt $ stand_power_x] && [$ stress_power lt $ stand_power_y] # judges that test node is whether in positive and negative 5% power consumption range to while

do

Stress_power=`ipmitool Ilanplus H192.168.1.2 Uadmin PadminTotal_Power`# band is outer regains up-to-date test node power consumption

The outer shutdown of ipmitool Ilanplus H192.168.1.2 Uadmin Padminpoweroff# band

Sleep20# adds time delay

The outer start of ipmitool Ilanplus H192.168.1.2 Uadmin Padminpoweron# band

done

Sleep60# adds time delay

After reboot# completes above action, control node is probably pressed and has been surveyed 120s, now restarts

D, entered in loop test from opening two test nodes of script by two above.

3) interpretation of result

Disk identifier of hard disk corresponding with slot log as follows:

By above it can be seen that start the hard disk SN that drive is corresponding every time, changing if there is hard disk drift or connection speed, then status will record error, very clear.

4) testing scheme assessment

The method of this cross validation is applicable to the hard disk compatibility test that a small amount of node carries out, and optimizes resource distribution, enriches test content, it is possible to make test machine repeating query test in different application scenarios, it is prevented that single test environment makes test machine produce inertia.It addition, this method is simple to operate, it is only necessary to test script is write start self-starting file/etc/rc.local, and test machine will be tested automatically, test can detect hard disk stability under multiple test condition mixes.

Claims (4)

1. one kind based on the method for dc and reboot cross validation hard disk stability under linux, it is characterised in that

The process of implementing is:

1) 2 node linux test environments, are built: according to the actual requirements, select different types of hard disk, SATA, SAS or SSD；

2), configuration systematic parameter: 2 test machine are installed operating system, installs RedHat, SUSE；

3), test environment configuration: require that 2 nodes open ipmi service, chkconfigipmion；Then row performs serviceipmistart, initializes ipmi state；

4), hard disk Detection of Stability: after preparation completes, by dc and reboot test that the node in dd pressure is hocketed, detect the slot position of hard disk, capacity, speed and smartctllog state after every time having tested.

2. method according to claim 1, it is characterised in that

Need powerstatus and Total_Power is judged at dc test phase, when test node having been performed be with outer poweroff order, need to go to obtain the powerstatus of node every 20s, when powerstatus is off, then to sending poweron order outside node band；If node powerstatus is on, controls node and will carry out dd stress test, after dd pressure surveys full 120s, perform reboot.

3. method according to claim 2, it is characterised in that

After test node completes OSinitialization after dc, dd stress test will be entered；And judge controlling node Total_Power, if within Total_Power value ± 5% that Total_Power value is when dd pressure time, node will be performed poweroff order；If Total_Power value is less than Total_Power value-5%, by sleep10, and then obtain Total_Power value and judge；

Each slot position detecting hard disk after being inducted into OS, capacity, speed and smartctllog state；Exception record is made mark, it is simple to the inspection after test.

4. method according to claim 3, it is characterised in that

The described step that realizes completes in being provided with the server of operating system of linux kernel, and during operation, user carries out with root identity logs.