LDP/LDP/howto/linuxdoc/Process-Monitor-HOWTO.sgml

1035 lines
33 KiB
Plaintext

<!doctype linuxdoc system>
<!--
************************** begin comment *****************************
The following is the HOW-TO for Monitoring Linux/Unix Processes.
This document is in the SGML format. You must use sgml package
to process this document
************************* end of comment *****************************
-->
<article>
<!-- Title information -->
<title>Process Monitor HOW-TO for Linux
<!-- chapt change
Process Monitor HOW-TO for Linux
-->
<author> Al Dev (Alavoor Vasudevan)
<htmlurl url="mailto:alavoor@yahoo.com"
name="alavoor@yahoo.com">
<date>v5.0, 21 April 2000
<abstract>
This document describes how to monitor Linux/Unix processes and to re-start them automatically
if they die without any manual intervention. This document also has URLs for "Unix Processes" FAQs.
</abstract>
<!-- Table of contents -->
<toc>
<!-- Begin the document -->
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> Linux or Unix Processes
-->
<sect> Linux or Unix Processes
<p>
Processes are the "heart" of the Linux/Unix processes. It is very important to monitor the
application processes to ensure 100% availability and reliability of the computer system.
For example, processes of databases, web-server etc.. need to be up and running 24 hours a
day and 365 days a year.
Use the tools described in this document to the monitor important application processes.
Also see the following related topics on Linux/Unix processes.
<itemize>
<item> Unix Programming FAQ - Chapter 1 Unix Processes <url url="http://www.erlenstar.demon.co.uk/unix/faq_toc.html">
<p>
<item> Other FAQs on Unix are at <url url="http://www.erlenstar.demon.co.uk/unix/">
<p>
</itemize>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> Unix/Linux command - procautostart
-->
<sect> Unix/Linux command - procautostart
<p>
Use the program <bf>procautostart</bf> (say "Prok-Auto-Start" or Process AutoStart) to
monitor and automatically re-start
any Unix/Linux process if they die. The program listing is given in following
sections in this document.
<bf>procautostart </bf> <bf>-n </bf> <it>&lt delay_seconds &gt</it> <bf>-c </bf> "<it>&lt command_line &gt</it>" nohup &amp
This starts the unix process <bf>procautostart</bf> and also <bf>command_line</bf>
process. The <bf>procautostart</bf> process will re-start <bf>command_line</bf>
process if it dies. The <it>-n</it> option is the time delay in seconds before <bf>procautostart</bf>
checks the running process started by <bf>command_line</bf>. It is advisable to start the procautostart as
background process with no-hangup using "nohup &amp". See 'man nohup'.
The procautostart is written in "C" so that it is very fast and efficient, since the program is called
every <it>n</it> seconds. Amount of resources consumed by procautostart is very minute.
For example -
<code>
procautostart -n 12 -c "monitor_test -d $HOME -a dummy_arg " nohup &amp
</code>
Here <bf>procautostart</bf> will be checking the process monitor_test <bf>every</bf> 12 seconds.
The program will output log files in 'mon' sub-directory which has datetime stamp of when the
processes died and re-started. These files gives info on how often the processes are dying.
You can also use micro-seconds option '-m' or nano-seconds option '-o', edit the source code file
<bf>procautostart.cpp</bf> and uncomment appropriate lines.
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> File procautostart.cpp
-->
<sect> File procautostart.cpp
<p>
// From your browser save this file as <bf>text-file</bf> named as 'procautostart.cpp'.
<code>
//
// Program to monitor the unix processes
// and automatically re-start them if they die
//
#include <stdio.h>
#include <strings.h> // C strings
#include <unistd.h> // for getopt
#include <alloc.h> // for free
#include <errno.h> // for kill() - error numbers command
extern int errno;
#ifdef Linux
#include <asm/errno.h> // for kill() - error numbers command
#endif
#include <sys/types.h> // for kill() command
#include <signal.h> // for kill() command
#include <sys/wait.h> // for wait()
#include <stdlib.h> // for setenv()
#include <time.h> // for strftime()
#include <libgen.h> // for basename()
#include "debug.h"
#define BUFF_HUN 100
#define BUFF_THOU 1024
#define PR_INIT_VAL -10
#define WAIT_FOR_SYS 5 // wait for process to start up
#define DEF_SL_SECS 6 // default sleep time
#define SAFE_MEM 10 // to avoid any possible memory leaks
#define LOG_NO false // do not output to logfile
#define LOG_YES true // do output to logfile
#define STD_ERR_NO false // do not print to std err
#define STD_ERR_YES true // do print to std err
#define DATE_NO false // do not print date
#define DATE_YES true // do print date
int start_process(char *commandline, char *args[], char **envp, pid_t proc_pid);
int fork2(pid_t parent_pid, unsigned long tsecs);
inline void error_msg(char *mesg_out, char *lg_file, bool pr_lg, bool std_err, bool pr_dt);
//////////////////////////////////////////////
// To test this program use --
// procautostart -n 5 -c 'monitor_test dummy1 -a dummy2 -b dummy3 ' &
//////////////////////////////////////////////
int main(int argc, char **argv, char **envp)
{
unsigned long sleep_sec, sleep_micro, sleep_nano;
int ch;
pid_t proc_pid;
int pr_no = PR_INIT_VAL;
char mon_log[40];
char *pr_name = NULL, *cmd_line = NULL, **cmdargs = NULL;
// you can turn on debug by editing Makefile and put -DDEBUG in gcc
debug_("test debug", "this line");
debug_("argc", argc);
// Use getpid() - man 2 getpid()
proc_pid = getpid(); // get the Process ID of procautostart
debug_("PID proc_pid", (int) proc_pid);
// Create directory to hold log, temp files
system("mkdir mon 1>/dev/null 2>/dev/null");
sleep_sec = DEF_SL_SECS ; // default sleep time
sleep_micro = 0; // default micro-sleep time
sleep_nano = 0; // default nano-sleep time
optarg = cmd_line = NULL;
while ((ch = getopt(argc, argv, "n:m:o:h:c:")) != -1) // needs trailing colon :
{
switch (ch)
{
case 'n':
debug_("scanned option n ", optarg);
sleep_sec = atoi(optarg);
debug_("sleep_sec", sleep_sec);
break;
case 'm':
debug_("scanned option m ", optarg);
sleep_micro = atoi(optarg);
debug_("sleep_micro", sleep_micro);
break;
case 'o':
debug_("scanned option o ", optarg);
sleep_nano = atoi(optarg);
debug_("sleep_nano", sleep_nano);
break;
case 'c':
debug_("scanned option c ", optarg);
cmd_line = strdup(optarg); // does auto-malloc here
debug_("cmd_line", cmd_line);
break;
case 'h':
debug_("scanned option h ", optarg);
fprintf(stderr, "\nUsage : %s -n <sleep> -m <microsecond> -o <nanosecond> -c '<command>'\n", argv[0]);
exit(-1);
break;
default:
debug_("ch", "default");
fprintf(stderr, "\nUsage : %s -n <sleep> -m <microsecond> -o <nanosecond> -c '<command>'\n", argv[0]);
exit(-1);
break;
}
}
if (cmd_line == NULL)
{
fprintf(stderr, "\ncmd_line is NULL");
fprintf(stderr, "\nUsage : %s -n <sleep> -m <microsecond> -o <nanosecond> -c '<command>'\n", argv[0]);
exit(-1);
}
else
{
// trim the trailing blanks -- otherwise problem in grep command
int tmpii = strlen(cmd_line);
for (int tmpjj = tmpii; tmpjj > -1; tmpjj--)
{
if (cmd_line[tmpjj] == ' ')
cmd_line[tmpjj] = '\0';
else
if (cmd_line[tmpjj] == '&') // discards amp-and .. we will be appending later
cmd_line[tmpjj] = '\0';
else
if (cmd_line[tmpjj] == '\0')
continue;
else
{
if (cmd_line[tmpjj] == '&') // Discard trailing & in command line
cmd_line[tmpjj] = '\0';
break;
}
}
debug_("cmd_line", cmd_line);
}
//argv0 = (char *) strdup(argv[0]);
//debug_("argv0", argv0);
// Start the process
{
// Find the command line args
char *aa = strdup(cmd_line), *bb = NULL;
cmdargs = (char **) (malloc(sizeof(char **) + SAFE_MEM));
for (int tmpii = 0; ; tmpii++)
{
// Allocate more memory ....
cmdargs = (char **) realloc(cmdargs, (sizeof(char **) * (tmpii+1) + SAFE_MEM) );
if (tmpii == 0)
bb = strtok(aa, " ");
else
bb = strtok(NULL, " "); // subsequent calls must have NULL as first arg
if (bb == NULL)
{
cmdargs[tmpii] = bb;
break;
}
else
{
// Must malloc with strdup because aa, bb are
// local vars in local scope!!
cmdargs[tmpii] = strdup(bb);
}
debug_("tmpii", tmpii);
debug_("cmdargs[tmpii]", (char *) cmdargs[tmpii]);
}
// In case execve you MUST NOT have trailing ampersand & in the command line!!
//pr_no = start_process(cmd_line, NULL, NULL, proc_pid); // Using execlp ...
pr_no = start_process(cmdargs[0], &amp cmdargs[0], envp, proc_pid); // Using execve ....
debug_("The child pid", pr_no);
if (pr_no < 0)
{
fprintf(stderr, "\nFatal Error: Failed to start the process\n");
exit(-1);
}
sleep(WAIT_FOR_SYS); // wait for the process to come up
// Get process name - only the first word from cmd_line
pr_name = strdup(basename(cmdargs[0])); // process name, does auto-malloc here
}
// generate log file names
{
char aa[21];
strncpy(aa, pr_name, 20); aa[20] = '\0';
// Define mon file-names - make it unique with combination of
// process name and process id
sprintf(mon_log, "mon/%s%d.log", aa, (int) proc_pid);
}
// Print out pid to log file
if (pr_no > 0)
{
char aa[200];
sprintf(aa, "Process ID of %s is %d", pr_name, pr_no);
error_msg(aa, mon_log, LOG_YES, STD_ERR_NO, DATE_YES);
}
// monitors the process - restarts if process dies...
bool process_died = false;
char print_log[200];
while (1) // infinite loop - monitor every 6 seconds
{
//debug_("Monitoring the process now...", ".");
if (kill(pr_no, 0)) // if (kill(pr_no,0) != 0)
{
debug_("errno from kill() function", errno);
if (errno == EINVAL)
{
process_died = false; // unable to execute kill() - wrong input
strcpy(print_log, "Error EINVAL: Invalid signal was specified");
error_msg(print_log, mon_log, LOG_YES, STD_ERR_YES, DATE_YES);
}
else
if (errno == ESRCH )
{
// ERSRCH means - No process can be found corresponding to pr_no
// hence process had died !!
process_died = true; // No process can be found matching pr_no
sprintf(print_log,
"Error ESRCH: No process or process group can be found for %d", pr_no);
error_msg(print_log, mon_log, LOG_YES, STD_ERR_YES, DATE_YES);
}
else
if (errno == EPERM)
{
process_died = false; // unable to execute kill() - wrong input
strcpy(print_log,
"Error EPERM: The real or saved user ID does not match the real user ID");
error_msg(print_log, mon_log, LOG_YES, STD_ERR_YES, DATE_YES);
}
else
{
process_died = true; // process died!! restart now
debug_("process_die ", "others");
}
if (process_died == true)
{
//
// char respawn[1024];
// strcpy(respawn, cmd_line);
//
// For "C" program use kill(pid_t process, int signal) function.
// #include <signal.h> // See 'man 2 kill'
// Returns 0 on success and -1 with errno set.
// kill -0 $pid 2>/dev/null || respawn
// To get the exit return status do --
// kill -0 $pid 2>/dev/null | echo $?
// Return value 0 is success and others mean failure
// Sending 0 does not do anything to target process, but it tests
// whether the process exists. The kill command will set its exit
// status based on this process.
//
// Alternatively, you can use
// ps -p $pid >/dev/null 2>&1 || respawn
// To get the exit return status do --
// ps -p $pid >/dev/null 2>&1 | echo $?
// Return value 0 is success and others mean failure
//
// If the process had died, restart and re-assign the pid to pr_no
// start the process in background ....
// Now re-assign new value of process id to pr_no
if (pr_no > 0 )
sprintf(print_log, "Fatal Error: Process %s with PID = %d died!!",
pr_name, pr_no);
else
sprintf(print_log, "Fatal Error: Process %s is not up!!",
pr_name);
error_msg(print_log, mon_log, LOG_YES, STD_ERR_YES, DATE_YES);
sprintf(print_log, "Starting process %s", pr_name);
error_msg(print_log, mon_log, LOG_YES, STD_ERR_NO, DATE_NO);
//pr_no = start_process(cmd_line, NULL, NULL, proc_pid); // Using execlp ....
pr_no = start_process(cmdargs[0], &amp cmdargs[0], envp, proc_pid); // Using execve ....
debug_("The child pid", pr_no);
if (pr_no < 0)
{
sprintf(print_log, "Fatal Error: Failed to start the process");
error_msg(print_log, mon_log, LOG_YES, STD_ERR_YES, DATE_YES);
exit(-1);
}
sleep(WAIT_FOR_SYS); // wait for the process to come up
sprintf(print_log, "Process ID of %s is %d", pr_name, pr_no);
error_msg(print_log, mon_log, LOG_YES, STD_ERR_NO, DATE_NO);
}
}
//debug_("Sleeping now ......", ".");
sleep(sleep_sec);
// Uncomment these to use micro-seconds
// For real-time process control use micro-seconds or nana-seconds sleep functions
// See 'man3 usleep', 'man 2 nanasleep'
// If you do not have usleep() or nanosleep() on your system, use select() or poll()
// specifying no file descriptors to test.
//usleep(sleep_micro);
// To sleep nano-seconds ... Uncomment these to use nano-seconds
//struct timespec *req = new struct timespec;
//req->tv_sec = 0; // seconds
//req->tv_nsec = sleep_nano; // nanoseconds
//nanosleep( (const struct timespec *)req, NULL);
}
}
inline void error_msg(char *mesg_out, char *lg_file, bool pr_lg, bool std_err, bool pr_dt)
{
if (pr_lg) // (pr_lg == true) output to log file
{
char tmp_msg[BUFF_THOU];
if (pr_dt == true) // print date and message to log file 'lg_file'
{
sprintf(tmp_msg, "date >> %s; echo '\n%s\n' >> %s\n ",
lg_file, mesg_out, lg_file);
system(tmp_msg);
}
else
{
sprintf(tmp_msg, "echo '\n%s\n' >> %s\n ",
mesg_out, lg_file);
system(tmp_msg);
}
}
if (std_err) // (std_err == true) output to standard error
fprintf(stderr, "\n%s\n", mesg_out);
debug_("mesg_out", mesg_out);
}
// start a process and returns PID or -ve value if error
// The main() function has envp arg as in - main(int argc, char *argv[], char **envp)
int start_process(char *commandline, char *args[], char **envp, pid_t parent_pid)
{
int ff;
unsigned long tsecs;
tsecs = time(NULL); // time in secs since Epoch 1 Jan 1970
debug_("Time tsecs", tsecs);
// Use fork2() instead of fork to avoid zombie child processes
switch (ff = fork2(parent_pid, tsecs)) // fork creates 2 process each executing the following lines
{
case -1:
fprintf(stderr, "\nFatal Error: start_process() - Unable to fork process\n");
_exit(errno);
break;
case 0: // child process
debug_("\nStarting the start child process\n", " ");
// For child process to ignore the interrupts (i.e. to put
// child process in "background" mode.
// Signals are sent to all processes started from a
// particular terminal. Accordingly, when a program is to be run non-interactively
// (started by &), the shell arranges that the program will ignore interrupts, so
// it won't be stopped by interrupts intended for foreground processes.
// Hence if previous value of signal is not IGN than set it to IGN.
// Note: Signal handlers cannot be set for SIGKILL, SIGSTOP
if (signal(SIGINT, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGINT\n");
else
if (signal(SIGINT, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGINT, SIG_IGN); // ignore interrupts
if (signal(SIGHUP, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGHUP\n");
else
if (signal(SIGHUP, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGHUP, SIG_IGN); // ignore hangups
if (signal(SIGQUIT, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGQUIT\n");
else
if (signal(SIGQUIT, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGQUIT, SIG_IGN); // ignore Quit
if (signal(SIGABRT, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGABRT\n");
else
if (signal(SIGABRT, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGABRT, SIG_IGN); // ignore ABRT
if (signal(SIGTERM, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGTERM\n");
else
if (signal(SIGTERM, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGTERM, SIG_IGN); // ignore TERM
// sigtstp - Stop typed at tty. Ignore this so that parent process
// be put in background with CTRL+Z or with SIGSTOP
if (signal(SIGTSTP, SIG_IGN) == SIG_ERR)
fprintf(stderr, "\nSignal Error: Not able to set signal to SIGTSTP\n");
else
if (signal(SIGTSTP, SIG_IGN) != SIG_IGN) // program already run in background
signal(SIGTSTP, SIG_IGN); // ignore TSTP
// You can use debug_ generously because they do NOT increase program size!
debug_("before execve commandline", commandline);
debug_("before execve args[0]", args[0]);
debug_("before execve args[1]", args[1]);
debug_("before execve args[2]", args[2]);
debug_("before execve args[3]", args[3]);
debug_("before execve args[4]", args[4]);
debug_("before execve args[5]", args[5]);
debug_("before execve args[6]", args[6]);
debug_("before execve args[7]", args[7]);
execve(commandline, args, envp);
// execlp, execvp does not provide expansion of metacharacters
// like <, >, *, quotes, etc., in argument list. Invoke
// the shell /bin/sh which then does all the work. Construct
// a string 'commandline' that contains the complete command
//execlp("/bin/sh", "sh", "-c", commandline, (char *) 0); // if success than NEVER returns !!
// If execlp returns than there is some serious error !! And
// executes the following lines below...
fprintf(stderr, "\nFatal Error: Unable to start child process\n");
ff = -2;
exit(127);
break;
default: // parent process
// child pid is ff;
if (ff < 0)
fprintf(stderr, "\nFatal Error: Problem while starting child process\n");
{
char buff[BUFF_HUN];
FILE *fp1;
sprintf(buff, "mon/%d%lu.out", (int) parent_pid, tsecs); // tsecs is unsigned long
fp1 = fopen(buff, "r");
if (fp1 != NULL)
{
buff[0] = '\0';
fgets(buff, BUFF_HUN, fp1);
ff = atoi(buff);
}
fclose(fp1);
debug_("start process(): ff - ", ff);
#ifndef DEBUG
sprintf(buff, "rm -f mon/%d%lu.out", (int) parent_pid, tsecs);
system(buff);
#endif // DEBUG
}
// define wait() to put child process in foreground or else put in background
//waitpid(ff, &amp status, WNOHANG || WUNTRACED);
//waitpid(ff, &amp status, WUNTRACED);
//wait(&amp status);
break;
}
return ff;
}
/* fork2() -- like fork, but the new process is immediately orphaned
* (won't leave a zombie when it exits)
* Returns 1 to the parent, not any meaningful pid.
* The parent cannot wait() for the new process (it's unrelated).
*/
/* This version assumes that you *haven't* caught or ignored SIGCHLD. */
/* If you have, then you should just be using fork() instead anyway. */
int fork2(pid_t parent_pid, unsigned long tsecs)
{
pid_t mainpid, child_pid = -10;
int status;
char buff[BUFF_HUN];
if (!(mainpid = fork()))
{
switch (child_pid = fork())
{
case 0:
//child_pid = getpid();
//debug_("At case 0 fork2 child_pid : ", child_pid);
return 0;
case -1:
_exit(errno); /* assumes all errnos are <256 */
default:
debug_("fork2 child_pid : ", (int) child_pid);
sprintf(buff, "echo %d > mon/%d%lu.out", (int) child_pid, (int) parent_pid, tsecs);
system(buff);
_exit(0);
}
}
//debug_("fork2 pid : ", pid);
if (mainpid < 0 || waitpid(mainpid, &amp status, 0) < 0)
return -1;
if (WIFEXITED(status))
if (WEXITSTATUS(status) == 0)
return 1;
else
errno = WEXITSTATUS(status);
else
errno = EINTR; /* well, sort of :-) */
return -1;
}
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> File debug.cpp
-->
<sect> File debug.cpp
<p>
// From your browser save this file as <bf>text-file</bf> named as 'debug.cpp'.
<p>
<code>
// This file defines the debug_() function which can be used for debugging
// the program. It is similar to "C" assert(). The debug_() is set to void()
// if DEBUG is not defined in Makefile. This way executable size of production
// release is NOT AT ALL effected. Using debug_() very generously has no
// impact on production executable size.
#ifdef DEBUG
#include "debug.h"
// Variable value[] can be char, string, int, unsigned long, float, etc...
void local_dbg(char name[], char value[], char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %s\n", fname, lineno, name, value ); }
void local_dbg(char name[], int value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], unsigned int value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], long value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], unsigned long value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], short value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], unsigned short value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %d\n", fname, lineno, name, value ); }
void local_dbg(char name[], float value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %f\n", fname, lineno, name, value ); }
void local_dbg(char name[], double value, char fname[], int lineno, bool logfile) {
printf("\nDebug %s Line: %d %s is = %f\n", fname, lineno, name, value ); }
// You add many more here - value can be a class, ENUM, datetime, etc...
#endif // DEBUG
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> File debug.h
-->
<sect> File debug.h
<p>
// From your browser save this file as <bf>text-file</bf> named as 'debug.h'.
<p>
<code>
#ifdef DEBUG
#include <stdio.h>
//#include <strings.h>
//#include <assert.h> // assert() macro which is also used for debugging
// Debugging code
// Use debug2_ to output result to a log file
#define debug_(NM, VL) (void) ( local_dbg(NM, VL, __FILE__, __LINE__) )
#define debug2_(NM, VL, LOG_FILE) (void) ( local_dbg(NM, VL, __FILE__, __LINE__, LOG_FILE) )
void local_dbg(char name[], char value[], char fname[], int lineno, bool logfile= false);
void local_dbg(char name[], int value, char fname[], int lineno, bool logfile= false);
void local_dbg(char name[], unsigned long value, char fname[], int lineno, bool logfile= false);
void local_dbg(char name[], float value, char fname[], int lineno, bool logfile= false);
#else
#define debug_(NM, VL) ((void) 0)
#define debug2_(NM, VL, LOG_FILE) ((void) 0)
#endif // DEBUG
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> Makefile
-->
<sect> Makefile
<p>
# From your browser save this file as <bf>text-file</bf> named as 'Makefile'.
<p>
<code>
EXE=procautostart
SRCS=procautostart.cpp debug.cpp
OBJS=procautostart.o debug.o
CXX=gcc
HOSTFLAG=-DLinux
#HOSTFLAG=-DSunOS
# Do not use compiler optimizer -O as this may break the program
# Use debug flag to enable the debug() function. If DEBUG is not
# defined than the function debug() is set to void(), similar
# to assert()
# Use options -Wall (all warning msgs) -O3 (optimization)
#MYCFLAGS=-DDEBUG -g -Wall
MYCFLAGS=-O3 -Wall
all: $(OBJS)
$(CXX) $(HOSTFLAG) $(MYCFLAGS) $(OBJS) -o $(EXE)
$(OBJS): $(SRCS)
$(CXX) -c $(HOSTFLAG) $(MYCFLAGS) $(SRCS)
clean:
rm -f *.o *.log *.log.old *.pid core err a.out afiedt.buf
rm -f $(EXE)
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> Testing the program - monitor_test
-->
<sect> Testing the program - monitor_test
<p>
From your browser save this file as <bf>text-file</bf> named as 'monitor_test'.
Use this program for testing the 'procautostart' program. For example -
<code>
procautostart -n 12 -c "monitor_test -d $HOME -a dummy_arg "
</code>
Here <bf>procautostart</bf> will be checking the process monitor_test <bf>every</bf> 12 seconds.
<p>
<code>
#!/bin/ksh
# Program to test the procautostart
echo "Started the monitor_test ...."
date > monitor_test.log
while :
do
date >> monitor_test.log
sleep 2
done
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt> Other Monitoring Tools
-->
<sect> Other Monitoring Tools
<p>
<sect1> OpenSource Monitoring Tools
<p>
On linux systems you can find the following packages. If it is not in the main
cdrom than you must check in the contrib cdrom :
<itemize>
<item> On contrib cdrom <bf>daemontools*.rpm</bf>
<p>
<item> 'top' command <bf>procps*.rpm</bf>
<p>
<item> 'top' command graphic mode <bf>procps-X11*.rpm</bf>
<p>
<item> 'ktop' graphic mode <bf>ktop*.rpm</bf>
<p>
<item> 'gtop' graphic mode <bf>gtop*.rpm</bf>
<p>
<item> 'WMMon' CPU load <bf>wmmon*.rpm</bf>
<p>
<item> 'wmsysmon' monitor <bf>wmsysmon*.rpm</bf>
<p>
<item> 'procmeter' System activity meter <bf>procmeter*.rpm</bf>
<p>
</itemize>
To use top commands type at unix prompt -
<code>
$ top
$ ktop
$ gtop
</code>
<!--
*******************************************
************ End of Section ***************
*******************************************
-->
<sect1> Monitoring Tool - "daemontools"
<p>
Visit the web site of daemontools at <url url="http://www.pobox.com/~djb/daemontools.html">
<p>
To install the daemontools RPM, do -
<code>
# rpm -i /mnt/cdrom/daemontools*.html
# man supervise
</code>
<bf>supervise</bf> monitors a service. It starts the service and restarts the
service if it dies. The companion svc program stops, pauses, or restarts
the service on sysadmin request. The svstat program prints a one-line
status report. See man page by 'man supervise'
<bf>svc</bf> - control a supervised service.
<p>
svc changes the status of a supervise-monitored service.
dir is the same directory used for supervise.
You can list several dirs. svc will change the status of
each service in turn.
<bf>svstat</bf> - print the status of a supervised service.
<p>
svstat prints the status of a supervise-monitored service.
dir is the same directory used for supervise.
You can list several dirs. svstat will print the status
of each service in turn.
<bf>cyclog</bf> writes a log to disk. It automatically synchronizes the log every
100KB (by default) to guarantee data integrity after a crash. It
automatically rotates the log to keep it below 1MB (by default). If the
disk fills up, cyclog pauses and then tries again, without losing any
data. See man page by 'man cyclog'
<bf>accustamp</bf> puts a precise timestamp on each line of input. The timestamp
is a numeric TAI timestamp with microsecond precision. The companion
tailocal program converts TAI timestamps to local time. See 'man accustamp'
<bf>usually</bf> watches a log for lines that do not match specified patterns,
copying those lines to stderr. The companion errorsto program redirects
stderr to a file. See 'man usually'
<bf>setuser</bf> runs a program under a user's uid and gid. Unlike su, setuser
does not gain privileges; it does not check passwords, and it cannot be
run except by root. See 'man setuser'
<!--
*******************************************
************ End of Section ***************
*******************************************
-->
<sect1> Commercial Monitoring Tools
<p>
There are commercial monitoring tools available. Check out -
<itemize>
<item> BMC Patrol for Unix/Databases <url url="http://www.bmc.com">
<p>
<item> TIBCO corp's Hawk for Unix monitoring <url url="http://www.tibco.com">
<p>
<item> LandMark corporation
<p>
<item> Platinum corporation
</itemize>
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt change> Other Formats of this Document
-->
<sect> Other Formats of this Document
<p>
This document is published in 11 different formats namely - DVI, Postscript,
Latex, Adobe Acrobat PDF,
LyX, GNU-info, HTML, RTF(Rich Text Format), Plain-text, Unix man pages and SGML.
<itemize>
<item>
You can get this HOWTO document as a single file tar ball in HTML, DVI,
Postscript or SGML formats from -
<url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO/other-formats/">
or
<url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO/other-formats/">
<item>Plain text format is in: <url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO">
or
<url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO">
<item>Translations to other languages like French, German, Spanish,
Chinese, Japanese are in
<url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO">
or <url url="ftp://metalab.unc.edu/pub/Linux/docs/HOWTO">
Any help from you to translate to other languages is welcome.
</itemize>
The document is written using a tool called "SGML tool" which can be got from -
<url url="http://www.xs4all.nl/~cg/sgmltools/">
Compiling the source you will get the following commands like
<itemize>
<item>sgml2html Process-Monitor-howto.sgml (to generate html file)
<item>sgml2rtf Process-Monitor-howto.sgml (to generate RTF file)
<item>sgml2latex Process-Monitor-howto.sgml (to generate latex file)
</itemize>
This document is located at -
<itemize>
<item> <url url="http://metalab.unc.edu/LDP/HOWTO/Process-Monitor-HOWTO.html">
</itemize>
Also you can find this document at the following mirrors sites -
<itemize>
<item> <url url="http://www.caldera.com/LDP/HOWTO/Process-Monitor-HOWTO.html">
<item> <url url="http://www.WGS.com/LDP/HOWTO/Process-Monitor-HOWTO.html">
<item> <url url="http://www.cc.gatech.edu/linux/LDP/HOWTO/Process-Monitor-HOWTO.html">
<item> <url url="http://www.redhat.com/linux-info/ldp/HOWTO/Process-Monitor-HOWTO.html">
<item> Other mirror sites near you (network-address-wise) can be found at
<url url="http://metalab.unc.edu/LDP/hmirrors.html">
select a site and go to directory /LDP/HOWTO/Process-Monitor-HOWTO.html
</itemize>
In order to view the document in dvi format, use the xdvi program. The xdvi
program is located in tetex-xdvi*.rpm package in Redhat Linux which can be
located through ControlPanel | Applications | Publishing | TeX menu buttons.
<tscreen><verb>
To read dvi document give the command -
xdvi -geometry 80x90 howto.dvi
And resize the window with mouse. See man page on xdvi.
To navigate use Arrow keys, Page Up, Page Down keys, also
you can use 'f', 'd', 'u', 'c', 'l', 'r', 'p', 'n' letter
keys to move up, down, center, next page, previous page etc.
To turn off expert menu press 'x'.
</verb></tscreen>
You can read postscript file using the program 'gv' (ghostview) or
'ghostscript'.
The ghostscript program is in ghostscript*.rpm package and gv
program is in gv*.rpm package in Redhat Linux
which can be located through ControlPanel | Applications | Graphics menu
buttons. The gv program is much more user friendly than ghostscript.
Ghostscript and gv are also available on other platforms like OS/2,
Windows 95 and NT.
<itemize>
<item>Get ghostscript for Windows 95, OS/2, and for all OSes from <url url="http://www.cs.wisc.edu/~ghost">
</itemize>
<tscreen><verb>
To read postscript document give the command -
gv howto.ps
To use ghostscript give -
ghostscript howto.ps
</verb></tscreen>
You can read HTML format document using Netscape Navigator, Microsoft Internet
explorer, Redhat Baron Web browser or any other web browsers.
You can read the latex, LyX output using LyX a "X-Windows" front end to latex.
<!--
*******************************************
************ End of Section ***************
*******************************************
<chapt change> Copyright Notice
-->
<sect> Copyright Notice
<p>
Copyright policy is GNU/GPL as per LDP (Linux Documentation project).
LDP is a GNU/GPL project.
Additional restrictions are - you must retain the author's name, email address
and this copyright notice on all the copies. If you make any changes
or additions to this document than you should
intimate all the authors of this document.
<!--
*******************************************
************ End of Section ***************
*******************************************
-->
</article>