Refactor check_load + new ouput

This commit is contained in:
Lorenz Kästle 2025-03-13 23:41:12 +01:00
parent f2900e0ccf
commit 08a475a14f
3 changed files with 273 additions and 176 deletions

View file

@ -56,6 +56,7 @@ EXTRA_DIST = t \
check_game.d \
check_radius.d \
check_time.d \
check_load.d \
check_nagios.d \
check_dbi.d \
check_tcp.d \

View file

@ -28,6 +28,9 @@
*
*****************************************************************************/
#include "output.h"
#include "perfdata.h"
#include "thresholds.h"
const char *progname = "check_load";
const char *copyright = "1999-2022";
const char *email = "devel@monitoring-plugins.org";
@ -36,6 +39,10 @@ const char *email = "devel@monitoring-plugins.org";
#include "./runcmd.h"
#include "./utils.h"
#include "./popen.h"
#include "states.h"
#include "check_load.d/config.h"
#include "../gl/stdlib.h"
#include <string.h>
@ -50,70 +57,68 @@ const char *email = "devel@monitoring-plugins.org";
# define LOADAVG_15MIN 2
#endif /* !defined LOADAVG_1MIN */
static int process_arguments(int argc, char **argv);
static int validate_arguments(void);
typedef struct {
int errorcode;
check_load_config config;
} check_load_config_wrapper;
static check_load_config_wrapper process_arguments(int argc, char **argv);
static check_load_config_wrapper validate_arguments(check_load_config_wrapper /*config_wrapper*/);
void print_help(void);
void print_usage(void);
static int print_top_consuming_processes();
typedef struct {
int errorcode;
char **top_processes;
} top_processes_result;
static top_processes_result print_top_consuming_processes(int /*n_procs_to_show*/);
static int n_procs_to_show = 0;
typedef struct {
mp_range load[3];
} parsed_thresholds;
static parsed_thresholds get_threshold(char *arg) {
size_t index;
char *str = arg;
char *tmp_pointer;
bool valid = false;
/* strictly for pretty-print usage in loops */
static const int nums[3] = {1, 5, 15};
parsed_thresholds result = {
.load[0] = mp_range_init(),
.load[1] = mp_range_init(),
.load[2] = mp_range_init(),
};
/* provide some fairly sane defaults */
double wload[3] = {0.0, 0.0, 0.0};
double cload[3] = {0.0, 0.0, 0.0};
#define la1 la[0]
#define la5 la[1]
#define la15 la[2]
char *status_line;
bool take_into_account_cpus = false;
static void get_threshold(char *arg, double *th) {
size_t i, n;
int valid = 0;
char *str = arg, *p;
n = strlen(arg);
for (i = 0; i < 3; i++) {
th[i] = strtod(str, &p);
if (p == str) {
size_t arg_length = strlen(arg);
for (index = 0; index < 3; index++) {
double tmp = strtod(str, &tmp_pointer);
if (tmp_pointer == str) {
break;
}
valid = 1;
str = p + 1;
if (n <= (size_t)(str - arg)) {
result.load[index] = mp_range_set_end(result.load[index], mp_create_pd_value(tmp));
valid = true;
str = tmp_pointer + 1;
if (arg_length <= (size_t)(str - arg)) {
break;
}
}
/* empty argument or non-floatish, so warn about it and die */
if (!i && !valid) {
if (!index && !valid) {
usage(_("Warning threshold must be float or float triplet!\n"));
}
if (i != 2) {
if (index != 2) {
/* one or more numbers were given, so fill array with last
* we got (most likely to NOT produce the least expected result) */
for (n = i; n < 3; n++) {
th[n] = th[i];
for (size_t tmp_index = index; tmp_index < 3; tmp_index++) {
result.load[tmp_index] = result.load[index];
}
}
return result;
}
int main(int argc, char **argv) {
int result = -1;
int i;
long numcpus;
double la[3] = {0.0, 0.0, 0.0}; /* NetBSD complains about uninitialized arrays */
#ifndef HAVE_GETLOADAVG
char input_buffer[MAX_INPUT_BUFFER];
#endif
setlocale(LC_ALL, "");
bindtextdomain(PACKAGE, LOCALEDIR);
textdomain(PACKAGE);
@ -122,112 +127,138 @@ int main(int argc, char **argv) {
/* Parse extra opts if any */
argv = np_extra_opts(&argc, argv, progname);
if (process_arguments(argc, argv) == ERROR) {
check_load_config_wrapper tmp_config = process_arguments(argc, argv);
if (tmp_config.errorcode == ERROR) {
usage4(_("Could not parse arguments"));
}
#ifdef HAVE_GETLOADAVG
result = getloadavg(la, 3);
if (result != 3) {
return STATE_UNKNOWN;
}
#else
child_process = spopen(PATH_TO_UPTIME);
if (child_process == NULL) {
printf(_("Error opening %s\n"), PATH_TO_UPTIME);
return STATE_UNKNOWN;
}
child_stderr = fdopen(child_stderr_array[fileno(child_process)], "r");
if (child_stderr == NULL) {
printf(_("Could not open stderr for %s\n"), PATH_TO_UPTIME);
}
fgets(input_buffer, MAX_INPUT_BUFFER - 1, child_process);
if (strstr(input_buffer, "load average:")) {
sscanf(input_buffer, "%*[^l]load average: %lf, %lf, %lf", &la1, &la5, &la15);
} else if (strstr(input_buffer, "load averages:")) {
sscanf(input_buffer, "%*[^l]load averages: %lf, %lf, %lf", &la1, &la5, &la15);
} else {
printf(_("could not parse load from uptime %s: %d\n"), PATH_TO_UPTIME, result);
return STATE_UNKNOWN;
const check_load_config config = tmp_config.config;
double load_values[3] = {0, 0, 0};
int error = getloadavg(load_values, 3);
if (error != 3) {
die(STATE_UNKNOWN, _("Failed to retrieve load values"));
}
result = spclose(child_process);
if (result) {
printf(_("Error code %d returned in %s\n"), result, PATH_TO_UPTIME);
return STATE_UNKNOWN;
}
#endif
if ((la[0] < 0.0) || (la[1] < 0.0) || (la[2] < 0.0)) {
#ifdef HAVE_GETLOADAVG
printf(_("Error in getloadavg()\n"));
#else
printf(_("Error processing %s\n"), PATH_TO_UPTIME);
#endif
return STATE_UNKNOWN;
mp_check overall = mp_check_init();
if (config.output_format_set) {
mp_set_format(config.output_format);
}
/* we got this far, so assume OK until we've measured */
result = STATE_OK;
xasprintf(&status_line, _("load average: %.2f, %.2f, %.2f"), la1, la5, la15);
xasprintf(&status_line, ("total %s"), status_line);
double scaled_la[3] = {0.0, 0.0, 0.0};
bool is_using_scaled_load_values = false;
if (take_into_account_cpus == true && (numcpus = GET_NUMBER_OF_CPUS()) > 0) {
long numcpus;
if (config.take_into_account_cpus && ((numcpus = GET_NUMBER_OF_CPUS()) > 0)) {
is_using_scaled_load_values = true;
scaled_la[0] = la[0] / numcpus;
scaled_la[1] = la[1] / numcpus;
scaled_la[2] = la[2] / numcpus;
double scaled_la[3] = {0.0, 0.0, 0.0};
scaled_la[0] = load_values[0] / numcpus;
scaled_la[1] = load_values[1] / numcpus;
scaled_la[2] = load_values[2] / numcpus;
char *tmp = NULL;
xasprintf(&tmp, _("load average: %.2f, %.2f, %.2f"), scaled_la[0], scaled_la[1], scaled_la[2]);
xasprintf(&status_line, "scaled %s - %s", tmp, status_line);
mp_subcheck scaled_load_sc = mp_subcheck_init();
scaled_load_sc = mp_set_subcheck_default_state(scaled_load_sc, STATE_OK);
scaled_load_sc.output = "Scaled Load (divided by number of CPUs";
mp_perfdata pd_scaled_load1 = perfdata_init();
pd_scaled_load1.label = "scaled_load1";
pd_scaled_load1 = mp_set_pd_value(pd_scaled_load1, scaled_la[0]);
pd_scaled_load1 = mp_pd_set_thresholds(pd_scaled_load1, config.th_load[0]);
mp_subcheck scaled_load_sc1 = mp_subcheck_init();
scaled_load_sc1 = mp_set_subcheck_state(scaled_load_sc1, mp_get_pd_status(pd_scaled_load1));
mp_add_perfdata_to_subcheck(&scaled_load_sc1, pd_scaled_load1);
xasprintf(&scaled_load_sc1.output, "1 Minute: %s", pd_value_to_string(pd_scaled_load1.value));
mp_add_subcheck_to_subcheck(&scaled_load_sc, scaled_load_sc1);
mp_perfdata pd_scaled_load5 = perfdata_init();
pd_scaled_load5.label = "scaled_load5";
pd_scaled_load5 = mp_set_pd_value(pd_scaled_load5, scaled_la[1]);
pd_scaled_load5 = mp_pd_set_thresholds(pd_scaled_load5, config.th_load[1]);
mp_subcheck scaled_load_sc5 = mp_subcheck_init();
scaled_load_sc5 = mp_set_subcheck_state(scaled_load_sc5, mp_get_pd_status(pd_scaled_load5));
mp_add_perfdata_to_subcheck(&scaled_load_sc5, pd_scaled_load5);
xasprintf(&scaled_load_sc5.output, "5 Minutes: %s", pd_value_to_string(pd_scaled_load5.value));
mp_add_subcheck_to_subcheck(&scaled_load_sc, scaled_load_sc5);
mp_perfdata pd_scaled_load15 = perfdata_init();
pd_scaled_load15.label = "scaled_load15";
pd_scaled_load15 = mp_set_pd_value(pd_scaled_load15, scaled_la[2]);
pd_scaled_load15 = mp_pd_set_thresholds(pd_scaled_load15, config.th_load[2]);
mp_subcheck scaled_load_sc15 = mp_subcheck_init();
scaled_load_sc15 = mp_set_subcheck_state(scaled_load_sc15, mp_get_pd_status(pd_scaled_load15));
mp_add_perfdata_to_subcheck(&scaled_load_sc15, pd_scaled_load15);
xasprintf(&scaled_load_sc15.output, "15 Minutes: %s", pd_value_to_string(pd_scaled_load15.value));
mp_add_subcheck_to_subcheck(&scaled_load_sc, scaled_load_sc15);
mp_add_subcheck_to_check(&overall, scaled_load_sc);
}
for (i = 0; i < 3; i++) {
if (is_using_scaled_load_values) {
if (scaled_la[i] > cload[i]) {
result = STATE_CRITICAL;
break;
} else if (scaled_la[i] > wload[i]) {
result = STATE_WARNING;
}
} else {
if (la[i] > cload[i]) {
result = STATE_CRITICAL;
break;
} else if (la[i] > wload[i]) {
result = STATE_WARNING;
mp_subcheck load_sc = mp_subcheck_init();
load_sc = mp_set_subcheck_default_state(load_sc, STATE_OK);
load_sc.output = "Total Load";
mp_perfdata pd_load1 = perfdata_init();
pd_load1.label = "load1";
pd_load1 = mp_set_pd_value(pd_load1, load_values[0]);
if (!is_using_scaled_load_values) {
pd_load1 = mp_pd_set_thresholds(pd_load1, config.th_load[0]);
}
mp_subcheck load_sc1 = mp_subcheck_init();
load_sc1 = mp_set_subcheck_state(load_sc1, mp_get_pd_status(pd_load1));
mp_add_perfdata_to_subcheck(&load_sc1, pd_load1);
xasprintf(&load_sc1.output, "1 Minute: %s", pd_value_to_string(pd_load1.value));
mp_add_subcheck_to_subcheck(&load_sc, load_sc1);
mp_perfdata pd_load5 = perfdata_init();
pd_load5.label = "load5";
pd_load5 = mp_set_pd_value(pd_load5, load_values[1]);
if (!is_using_scaled_load_values) {
pd_load5 = mp_pd_set_thresholds(pd_load5, config.th_load[1]);
}
mp_subcheck load_sc5 = mp_subcheck_init();
load_sc5 = mp_set_subcheck_state(load_sc5, mp_get_pd_status(pd_load5));
mp_add_perfdata_to_subcheck(&load_sc5, pd_load5);
xasprintf(&load_sc5.output, "5 Minutes: %s", pd_value_to_string(pd_load5.value));
mp_add_subcheck_to_subcheck(&load_sc, load_sc5);
mp_perfdata pd_load15 = perfdata_init();
pd_load15.label = "load15";
pd_load15 = mp_set_pd_value(pd_load15, load_values[2]);
if (!is_using_scaled_load_values) {
pd_load15 = mp_pd_set_thresholds(pd_load15, config.th_load[2]);
}
mp_subcheck load_sc15 = mp_subcheck_init();
load_sc15 = mp_set_subcheck_state(load_sc15, mp_get_pd_status(pd_load15));
mp_add_perfdata_to_subcheck(&load_sc15, pd_load15);
xasprintf(&load_sc15.output, "15 Minutes: %s", pd_value_to_string(pd_load15.value));
mp_add_subcheck_to_subcheck(&load_sc, load_sc15);
mp_add_subcheck_to_check(&overall, load_sc);
if (config.n_procs_to_show > 0) {
mp_subcheck top_proc_sc = mp_subcheck_init();
top_proc_sc = mp_set_subcheck_state(top_proc_sc, STATE_OK);
top_processes_result top_proc = print_top_consuming_processes(config.n_procs_to_show);
top_proc_sc.output = "";
if (top_proc.errorcode == OK) {
for (int i = 0; i < config.n_procs_to_show; i++) {
xasprintf(&top_proc_sc.output, "%s\n%s", top_proc_sc.output, top_proc.top_processes[i]);
}
}
}
printf("LOAD %s - %s|", state_text(result), status_line);
for (i = 0; i < 3; i++) {
if (is_using_scaled_load_values) {
printf("load%d=%.3f;;;0; ", nums[i], la[i]);
printf("scaled_load%d=%.3f;%.3f;%.3f;0; ", nums[i], scaled_la[i], wload[i], cload[i]);
} else {
printf("load%d=%.3f;%.3f;%.3f;0; ", nums[i], la[i], wload[i], cload[i]);
}
}
putchar('\n');
if (n_procs_to_show > 0) {
print_top_consuming_processes();
}
return result;
mp_exit(overall);
}
/* process command-line arguments */
static int process_arguments(int argc, char **argv) {
int c = 0;
int option = 0;
static check_load_config_wrapper process_arguments(int argc, char **argv) {
static struct option longopts[] = {{"warning", required_argument, 0, 'w'},
{"critical", required_argument, 0, 'c'},
{"percpu", no_argument, 0, 'r'},
@ -236,26 +267,45 @@ static int process_arguments(int argc, char **argv) {
{"procs-to-show", required_argument, 0, 'n'},
{0, 0, 0, 0}};
check_load_config_wrapper result = {
.errorcode = OK,
.config = check_load_config_init(),
};
if (argc < 2) {
return ERROR;
result.errorcode = ERROR;
return result;
}
while (1) {
c = getopt_long(argc, argv, "Vhrc:w:n:", longopts, &option);
while (true) {
int option = 0;
int option_index = getopt_long(argc, argv, "Vhrc:w:n:", longopts, &option);
if (c == -1 || c == EOF) {
if (option_index == -1 || option_index == EOF) {
break;
}
switch (c) {
case 'w': /* warning time threshold */
get_threshold(optarg, wload);
break;
case 'c': /* critical time threshold */
get_threshold(optarg, cload);
break;
switch (option_index) {
case 'w': /* warning time threshold */ {
parsed_thresholds warning_range = get_threshold(optarg);
result.config.th_load[0].warning = warning_range.load[0];
result.config.th_load[0].warning_is_set = true;
result.config.th_load[1].warning = warning_range.load[1];
result.config.th_load[1].warning_is_set = true;
result.config.th_load[2].warning = warning_range.load[2];
result.config.th_load[2].warning_is_set = true;
} break;
case 'c': /* critical time threshold */ {
parsed_thresholds critical_range = get_threshold(optarg);
result.config.th_load[0].critical = critical_range.load[0];
result.config.th_load[0].critical_is_set = true;
result.config.th_load[1].critical = critical_range.load[1];
result.config.th_load[1].critical_is_set = true;
result.config.th_load[2].critical = critical_range.load[2];
result.config.th_load[2].critical_is_set = true;
} break;
case 'r': /* Divide load average by number of CPUs */
take_into_account_cpus = true;
result.config.take_into_account_cpus = true;
break;
case 'V': /* version */
print_revision(progname, NP_VERSION);
@ -264,49 +314,49 @@ static int process_arguments(int argc, char **argv) {
print_help();
exit(STATE_UNKNOWN);
case 'n':
n_procs_to_show = atoi(optarg);
result.config.n_procs_to_show = atoi(optarg);
break;
case '?': /* help */
usage5();
}
}
c = optind;
if (c == argc) {
return validate_arguments();
int index = optind;
if (index == argc) {
return validate_arguments(result);
}
/* handle the case if both arguments are missing,
* but not if only one is given without -c or -w flag */
if (c - argc == 2) {
get_threshold(argv[c++], wload);
get_threshold(argv[c++], cload);
} else if (c - argc == 1) {
get_threshold(argv[c++], cload);
if (index - argc == 2) {
parsed_thresholds warning_range = get_threshold(argv[index++]);
result.config.th_load[0].warning = warning_range.load[0];
result.config.th_load[0].warning_is_set = true;
result.config.th_load[1].warning = warning_range.load[1];
result.config.th_load[1].warning_is_set = true;
result.config.th_load[2].warning = warning_range.load[2];
result.config.th_load[2].warning_is_set = true;
parsed_thresholds critical_range = get_threshold(argv[index++]);
result.config.th_load[0].critical = critical_range.load[0];
result.config.th_load[0].critical_is_set = true;
result.config.th_load[1].critical = critical_range.load[1];
result.config.th_load[1].critical_is_set = true;
result.config.th_load[2].critical = critical_range.load[2];
result.config.th_load[2].critical_is_set = true;
} else if (index - argc == 1) {
parsed_thresholds critical_range = get_threshold(argv[index++]);
result.config.th_load[0].critical = critical_range.load[0];
result.config.th_load[0].critical_is_set = true;
result.config.th_load[1].critical = critical_range.load[1];
result.config.th_load[1].critical_is_set = true;
result.config.th_load[2].critical = critical_range.load[2];
result.config.th_load[2].critical_is_set = true;
}
return validate_arguments();
return validate_arguments(result);
}
static int validate_arguments(void) {
int i = 0;
/* match cload first, as it will give the most friendly error message
* if user hasn't given the -c switch properly */
for (i = 0; i < 3; i++) {
if (cload[i] < 0) {
die(STATE_UNKNOWN, _("Critical threshold for %d-minute load average is not specified\n"), nums[i]);
}
if (wload[i] < 0) {
die(STATE_UNKNOWN, _("Warning threshold for %d-minute load average is not specified\n"), nums[i]);
}
if (wload[i] > cload[i]) {
die(STATE_UNKNOWN, _("Parameter inconsistency: %d-minute \"warning load\" is greater than \"critical load\"\n"), nums[i]);
}
}
return OK;
}
static check_load_config_wrapper validate_arguments(check_load_config_wrapper config_wrapper) { return config_wrapper; }
void print_help(void) {
print_revision(progname, NP_VERSION);
@ -363,23 +413,39 @@ int cmpstringp(const void *p1, const void *p2) {
}
#endif /* PS_USES_PROCPCPU */
static int print_top_consuming_processes() {
int i = 0;
struct output chld_out, chld_err;
static top_processes_result print_top_consuming_processes(int n_procs_to_show) {
top_processes_result result = {
.errorcode = OK,
};
struct output chld_out;
struct output chld_err;
if (np_runcmd(PS_COMMAND, &chld_out, &chld_err, 0) != 0) {
fprintf(stderr, _("'%s' exited with non-zero status.\n"), PS_COMMAND);
return STATE_UNKNOWN;
result.errorcode = ERROR;
return result;
}
if (chld_out.lines < 2) {
fprintf(stderr, _("some error occurred getting procs list.\n"));
return STATE_UNKNOWN;
result.errorcode = ERROR;
return result;
}
#ifdef PS_USES_PROCPCPU
qsort(chld_out.line + 1, chld_out.lines - 1, sizeof(char *), cmpstringp);
#endif /* PS_USES_PROCPCPU */
int lines_to_show = chld_out.lines < (size_t)(n_procs_to_show + 1) ? (int)chld_out.lines : n_procs_to_show + 1;
for (i = 0; i < lines_to_show; i += 1) {
printf("%s\n", chld_out.line[i]);
result.top_processes = calloc(lines_to_show, sizeof(char *));
if (result.top_processes == NULL) {
// Failed allocation
result.errorcode = ERROR;
return result;
}
return OK;
for (int i = 0; i < lines_to_show; i += 1) {
xasprintf(&result.top_processes[i], "%s", chld_out.line[i]);
}
return result;
}

View file

@ -0,0 +1,30 @@
#pragma once
#include "output.h"
#include "thresholds.h"
typedef struct {
mp_thresholds th_load[3];
bool take_into_account_cpus;
int n_procs_to_show;
mp_output_format output_format;
bool output_format_set;
} check_load_config;
check_load_config check_load_config_init() {
check_load_config tmp = {
.th_load =
{
mp_thresholds_init(),
mp_thresholds_init(),
mp_thresholds_init(),
},
.take_into_account_cpus = false,
.n_procs_to_show = 0,
.output_format_set = false,
};
return tmp;
}