CN101630262A

CN101630262A - Method for monitoring and controlling subprocess based on Linux system

Info

Publication number: CN101630262A
Application number: CN200910089410A
Authority: CN
Inventors: 裴文浩
Original assignee: BEIJING SHUSHUAI TECHNOLOGY Co Ltd
Current assignee: BEIJING SHUSHUAI TECHNOLOGY Co Ltd
Priority date: 2009-07-17
Filing date: 2009-07-17
Publication date: 2010-01-20

Abstract

The invention provides a method for monitoring and controlling a subprocess based on a Linux system, comprising the following steps of: after starting up an operation system, loading a task list, and executing tasks in the task list; building a pid document which comprises a distributed pid for the subprocess which needs to be ministered and controlled; obtaining the pid of the subprocess when an exit signal SIGCHLD of the subprocess is judged to be received; obtaining a pid document name corresponding to the pid of the subprocess when the obtained pid of the subprocess is judged to be accordance with the pid in the built pid document; obtaining a control script name in a restarted configuration document corresponding to the pid document; and executing the control script corresponding to the control script name, and generating and executing a task for restarting the subprocess. The method for monitoring and controlling the subprocess based on the Linux system can monitor and control the quit of the subprocess, realize restart of the quit subprocess, and save system resource.

Description

Subprocess method for supervising based on (SuSE) Linux OS

Technical field

The present invention relates to the process monitoring field, particularly relate to a kind of subprocess method for supervising based on (SuSE) Linux OS.

Background technology

The service subprocess is to have the program of specific function rather than the ingredient of linux kernel.The service subprocess when system bootstrap, start usually, and when (SuSE) Linux OS moves continuous service.The Init process is first process that will move after the system bootstrap, is the parent process of all service subprocesss.Common service subprocess comprises the process of sealing, file processing subprocess, management database subprocess, Email subprocess and network subprocess etc.Running into unknown problem or mistake in system's operational process can cause these service subprocess accidents to withdraw from.At this moment need artificially to start the service subprocess that these withdraw from, otherwise the process that withdraws from can not be resumed operation.

Now, the following method of general employing addresses the above problem: monitoring process is set, this monitoring process adopts the mode of uninterrupted poll to check whether the service subprocess moves, check out that a certain service subprocess does not have operation just to restart this service subprocess if monitoring process is set, idiographic flow as shown in Figure 1.Though this method can solve the problem that service subprocess accident withdraws from, adopt monitoring process that each process is carried out uninterrupted poll check and not only can expend a large amount of system resource, but also can not realize starting in real time the unexpected service subprocess that withdraws from.

Summary of the invention

In view of this, the invention provides a kind of subprocess method for supervising, can withdrawing from of subprocess be monitored, realize withdrawing from restarting of subprocess, save system resource based on (SuSE) Linux OS.

Subprocess method for supervising based on (SuSE) Linux OS provided by the invention, behind the os starting, the task in the described task list is carried out in the loading tasks tabulation, also comprises step:

A, set up the pid file of the record pid that this subprocess distributed for the subprocess that needs monitoring;

B, judgement receive the pid that obtains described subprocess when subprocess withdraws from signal SIGCHLD;

The pid that puts down in writing in the pid of C, more obtained subprocess and the pid file of foundation obtains the title of this pid file when the two is consistent;

D, obtain the control script name that restarts the described pid file name correspondence that writes down in the configuration file;

The control script of E, the described control script name correspondence of execution generates the task of restarting subprocess of also carrying out.

As from the foregoing, adopt above-mentioned steps to monitor, realize withdrawing from restarting of subprocess,, save system resource, and can realize real-time restarting with respect to the background technology polling mode to withdrawing from of subprocess.

Wherein, described steps A comprises substep:

The pid file of the subprocess correspondence that will monitor that A1, foundation start;

A2, obtain the pid of current promoter process;

A3, described pid is write corresponding pid file.

As from the foregoing, the employing above-mentioned steps can be embodied as the pid file that the subprocess foundation that need monitor comprises the pid that is distributed fast.

Wherein, the described judgement of step B receives subprocess and withdraws from signal SIGCHLD and judged by the init process;

The described pid that obtains described subprocess is obtained by the wait system function of init process transfer.

As from the foregoing, adopt above-mentioned steps can monitor the situation that withdraws from of subprocess in real time.

Wherein, the pid file storage set up of steps A is under/var/run catalogue;

Step C comprises substep:

The pid that each pid file under the pid of obtained subprocess and the/var/run catalogue is comprised compares;

When comparative result is identical, obtain described pid file name.

As from the foregoing, adopt above-mentioned steps can conveniently obtain pid for the pid file name.

Wherein, step D is described to restart configuration file stores under/etc/init.d catalogue, comprises following field at least:

Pid file name, control script name.

As from the foregoing, adopt the above-mentioned configuration file that restarts that the subprocess that need monitor can be set flexibly.

Wherein, the described task of becoming to restart of carrying out of step e comprises the substep that generates new pid file:

The pid file of E1, deletion pid file name correspondence; The pid file of the subprocess correspondence that E2, foundation restart; E3, obtain the pid of current promoter process; E4, described pid is write corresponding pid file.

As from the foregoing, adopt above-mentioned steps can delete the pid file that withdraws from subprocess, set up new pid file, help the subprocess that restarts is withdrawed from monitoring for the subprocess that restarts.

As from the foregoing, adopt above-mentioned steps to be convenient to carry out fast the task that restarts.

Preferable, also comprise after the step e: destroy the described subprocess that withdraws from according to the control script.

As from the foregoing, adopt above-mentioned steps can reduce the unnecessary wasting of resources, improve system running speed.

Description of drawings

Fig. 1 is the process flow diagram of available technology adopting monitoring process poll monitor service subprocess;

Fig. 2 is the process flow diagram of the subprocess method for supervising based on (SuSE) Linux OS provided by the invention;

Fig. 3 is the structural representation that restarts configuration file;

Fig. 4 is the pid file, restart the corresponding relation synoptic diagram of configuration file and control pin presents.

Embodiment

The flow process of the subprocess method for supervising based on (SuSE) Linux OS provided by the invention as shown in Figure 2, comprises step:

Step 201, start-up system (beginning).

Step 202, system carry out the initialization setting, comprise the initialization control desk, terminal attribute is set, system environment variable is set, operation such as shielding and adapter signal.

Step 203, parsing start configuration file inittab, and need moving when loading the init process initiation of task generates task list.The task structure body is as follows:

struct?init_action{

Struct init_action*next; Single linked list next pointer

Int action; The current task type

Pid_t pid; Current task pid

Char command[INIT_BUFFS_SIZE]; Current task command execution complete trails

Char terminal[CONSOLE_BUFF_SIZE]; The output control desk of current task

}。

Task in step 204, the tabulation of executing the task starts each subprocess, generate the pid file of the subprocess correspondence that needs monitor according to the control script, and be based upon/the var/run catalogue under.

Wherein, the method for building up of pid file is: obtain the pid (Process identifier) that distributes to the subprocess of current startup by calling pidof (order of linux system standard), pid is write corresponding pid file.The naming rule of pid file is self-defined title (normally task names) .pid (the suffix name of " pid " expression file in this place's naming rule).For example the subprocess of current startup is a ftp service subprocess, and the pid that distributes to ftp service subprocess when it starts is 843, and then the pid file name of this subprocess is ftp.pid, and the content of file is 843.If subprocess is the multi-process service, then the pid with each process is kept in the same pid file with space-separated.

Executed task in step 205, the deletion task list.Can avoid repeating of task like this, save system resource.

Step 206, init process wait for that sub-journey withdraws from, and when the anyon process withdrawed from, the init process will receive SIGCHLD signal (the SIGCHLD signal is that subprocess withdraws from signal, when being the subprocess end, inwardly authorizes the signal that send).Withdraw from signal if receive subprocess, enter step 207, otherwise repeating step 206.

Step 207, call the wait system function, obtain the described pid that withdraws from subprocess by the wait system function.

Pid in pid that withdraws from subprocess that step 208, comparison wait system function are obtained and the pid listed files in each pid file.The pid listed files is kept at/the var/run catalogue under, as shown in Figure 4.Withdraw from if both pid unanimities, explanation are the subprocesss of setting up current pid file, enter step 209.If inconsistent, enter step 206.

Step 209, record withdraw from the pid file name of the pid correspondence of subprocess.

Step 210, by line by line relatively the title of pid file and the critical field 1 that restarts configuration file obtain the control script name of the title correspondence of pid file, promptly restart the critical field 3 of configuration file.Restart the profile list be stored in usually/the etc/init.d catalogue under, show the structure that restarts configuration file as Fig. 3, comprise pid file name (critical field 1): function expansion word (critical field 2): control script name (critical field 3).Fig. 4 shows the relation that restarts configuration file, pid listed files, script control tabulation.

The control script of the control script name correspondence that step 211, execution are obtained generates restarting of task, and executing the task generates new subprocess, rebuilds corresponding pid file.The task structure body that restarts is as follows:

struct?init_action{

Struct init_action*next; Single linked list next pointer

Int action; The current task type

Pid_tpid; Current task pid

Char command[INIT_BUFFS_SIZE]; Current task command execution complete trails

Char termina l[CONSOLE_BUFF_SIZE]; The output control desk of current task

}；

Wherein,

Next member is NULL;

Action member is ONCE;

Pid member is 0;

Command member is control script complete trails/script name.With the ntp subprocess is that its control script complete trails of example is :/etc/init.d/ntp.sh

Terminal member is the subprocess control desk.Be defaulted as/dev/console

Wherein, rebuild corresponding pid file and comprise step: 1, the pid file of deletion pid file name correspondence; 2, set up the pid file of the subprocess correspondence that restarts; 3, obtain the pid of current promoter process; 4, described pid is write corresponding pid file.

Step 212, destroy according to the control script and to withdraw from subprocess, discharge and withdraw from the shared resource of subprocess.It is for fear of causing the unnecessary wasting of resources that this step reclaims resource.

Among the present invention each need withdraw from corresponding one of the subprocess of monitoring and independently control script.Control script be generally held in/the etc catalogue under.The control script comprises three functional modules: start functional module, exit function module and restart function module.Call the startup functional module in the control script in the time of system start-up, call the restart function module in the control script when restarting.

Below be the compiling procedure of the control script of example explanation ntp subprocess with the ntp subprocess.

Step 1 is determined the pid file name of ntp task: ntp.pid and control script name: ntp.sh.

Step 2, the function of writing ntp control script.The control example script is as follows:

#！/bin/sh

#description：Starts?and?stops?the?ntp?daemons?\

# used?to?provide?ntp?services.

#

#pidfile：/var/run/ntp.pid

#Const

PIDPATH＝/var/run

PRGNAME＝ntp

SRVNAME＝ntp

#Source?function?library.

./etc/init.d/vstfunc

#Check?that?ntp.conf?exists.

[-f/etc/ntp/ntp.cfg]||exit?0

#Function

start(){

#check?program

checkonly?$PRGNAME

if?[$？-eq?0]；then

exit?0

fi

sleep?3

/usr/sbin/$PRGNAME

sleep?3

return$？

}

stop(){

killproc?$PRGNAME

return?$？

}

restart(){

stop

start

}

#?Select

case″$1″in

start)

start

；；

stop)

stop

；；

restart)

restart

；；

*)

echo?$″Usage：$0{start|stop|restart}″

exit?1

esac

exit?$？

Add ntp::ntp.sh to restarting under the configuration file catalogue, can the ntp subprocess be set to the subprocess that to monitor.

The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims

1. subprocess method for supervising based on (SuSE) Linux OS, behind the os starting, the task in the described task list is carried out in the loading tasks tabulation, it is characterized in that, also comprises step:

2. method according to claim 1 is characterized in that, described steps A comprises substep:

A2, obtain the pid of the described subprocess of current startup;

A3, described pid is write corresponding pid file.

3. method according to claim 1 is characterized in that, the described judgement of step B receives subprocess and withdraws from signal SIGCHLD and judged by the init process;

4. method according to claim 1 is characterized in that,

The pid file storage that steps A is set up is under/var/run catalogue;

Step C comprises substep:

When comparative result is identical, obtain described pid file name.

5. according to claim 1 or 4 described methods, it is characterized in that step D is described to restart configuration file stores under/etc/init.d catalogue, comprises following field at least:

Pid file name, control script name.

6. method according to claim 1 is characterized in that, the task that the described execution of step e restarts subprocess comprises the substep that generates new pid file:

The pid file of E1, deletion pid file name correspondence;

The pid file of the subprocess correspondence that E2, foundation restart;

E3, obtain the pid of the current subprocess that restarts;

E4, described pid is write corresponding pid file.

7. method according to claim 6 is characterized in that, also comprises after the step e:

Destroy the described subprocess that withdraws from according to the control script.