* Copyright (C) 2009 Andrés J. Díaz
* Copyright (C) 2009 Manuel Sanmartin
* Copyright (C) 2010 Clément Stenac
+ * Copyright (C) 2012 Cosmin Ioiart
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the
* Andrés J. Díaz <ajdiaz at connectical.com>
* Manuel Sanmartin
* Clément Stenac <clement.stenac at diwi.org>
+ * Cosmin Ioiart <cioiart at gmail.com>
**/
#include "collectd.h"
#define MAXARGLN 1024
/* #endif HAVE_PROCINFO_H */
+#elif KERNEL_SOLARIS
+# include <procfs.h>
+# include <dirent.h>
+/* #endif KERNEL_SOLARIS */
+
#else
# error "No applicable input method."
#endif
# include <regex.h>
#endif
+#if HAVE_KSTAT_H
+# include <kstat.h>
+#endif
+
#ifndef ARG_MAX
# define ARG_MAX 4096
#endif
/* #endif KERNEL_LINUX */
#elif HAVE_LIBKVM_GETPROCS && HAVE_STRUCT_KINFO_PROC_FREEBSD
-/* no global variables */
+static int pagesize;
/* #endif HAVE_LIBKVM_GETPROCS && HAVE_STRUCT_KINFO_PROC_FREEBSD */
#elif HAVE_PROCINFO_H
#endif /* HAVE_PROCINFO_H */
/* put name of process from config to list_head_g tree
- list_head_g is a list of 'procstat_t' structs with
- processes names we want to watch */
+ * list_head_g is a list of 'procstat_t' structs with
+ * processes names we want to watch */
static void ps_list_register (const char *name, const char *regexp)
{
procstat_t *new;
/* #endif KERNEL_LINUX */
#elif HAVE_LIBKVM_GETPROCS && HAVE_STRUCT_KINFO_PROC_FREEBSD
-/* no initialization */
+ pagesize = getpagesize();
/* #endif HAVE_LIBKVM_GETPROCS && HAVE_STRUCT_KINFO_PROC_FREEBSD */
#elif HAVE_PROCINFO_H
}
DEBUG ("name = %s; num_proc = %lu; num_lwp = %lu; "
- "vmem_size = %lu; vmem_rss = %lu; vmem_data = %lu; "
+ "vmem_size = %lu; vmem_rss = %lu; vmem_data = %lu; "
"vmem_code = %lu; "
"vmem_minflt_counter = %"PRIi64"; vmem_majflt_counter = %"PRIi64"; "
"cpu_user_counter = %"PRIi64"; cpu_system_counter = %"PRIi64"; "
ps->io_rchar, ps->io_wchar, ps->io_syscr, ps->io_syscw);
} /* void ps_submit_proc_list */
+#if KERNEL_LINUX || KERNEL_SOLARIS
+static void ps_submit_fork_rate (derive_t value)
+{
+ value_t values[1];
+ value_list_t vl = VALUE_LIST_INIT;
+
+ values[0].derive = value;
+
+ vl.values = values;
+ vl.values_len = 1;
+ sstrncpy(vl.host, hostname_g, sizeof (vl.host));
+ sstrncpy(vl.plugin, "processes", sizeof (vl.plugin));
+ sstrncpy(vl.plugin_instance, "", sizeof (vl.plugin_instance));
+ sstrncpy(vl.type, "fork_rate", sizeof (vl.type));
+ sstrncpy(vl.type_instance, "", sizeof (vl.type_instance));
+
+ plugin_dispatch_values(&vl);
+}
+#endif /* KERNEL_LINUX || KERNEL_SOLARIS*/
+
/* ------- additional functions for KERNEL_LINUX/HAVE_THREAD_INFO ------- */
#if KERNEL_LINUX
static int ps_read_tasks (int pid)
continue;
numfields = strsplit (buffer, fields,
- STATIC_ARRAY_SIZE (fields));
+ STATIC_ARRAY_SIZE (fields));
if (numfields < 2)
continue;
tmp = strtoll (fields[1], &endptr, /* base = */ 10);
if ((errno == 0) && (endptr != fields[1]))
{
- if (strncmp (buffer, "VmData", 6) == 0)
+ if (strncmp (buffer, "VmData", 6) == 0)
{
data = tmp;
}
return buf;
} /* char *ps_get_cmdline (...) */
-static unsigned long read_fork_rate ()
+static int read_fork_rate ()
{
FILE *proc_stat;
- char buf[1024];
- unsigned long result = 0;
- int numfields;
- char *fields[3];
+ char buffer[1024];
+ value_t value;
+ _Bool value_valid = 0;
- proc_stat = fopen("/proc/stat", "r");
- if (proc_stat == NULL) {
+ proc_stat = fopen ("/proc/stat", "r");
+ if (proc_stat == NULL)
+ {
char errbuf[1024];
ERROR ("processes plugin: fopen (/proc/stat) failed: %s",
sstrerror (errno, errbuf, sizeof (errbuf)));
- return ULONG_MAX;
+ return (-1);
}
- while (fgets (buf, sizeof(buf), proc_stat) != NULL)
+ while (fgets (buffer, sizeof (buffer), proc_stat) != NULL)
{
- char *endptr;
+ int status;
+ char *fields[3];
+ int fields_num;
- numfields = strsplit(buf, fields, STATIC_ARRAY_SIZE (fields));
- if (numfields != 2)
+ fields_num = strsplit (buffer, fields,
+ STATIC_ARRAY_SIZE (fields));
+ if (fields_num != 2)
continue;
if (strcmp ("processes", fields[0]) != 0)
continue;
- errno = 0;
- endptr = NULL;
- result = strtoul(fields[1], &endptr, /* base = */ 10);
- if ((endptr == fields[1]) || (errno != 0)) {
- ERROR ("processes plugin: Cannot parse fork rate: %s",
- fields[1]);
- result = ULONG_MAX;
- break;
- }
+ status = parse_value (fields[1], &value, DS_TYPE_DERIVE);
+ if (status == 0)
+ value_valid = 1;
break;
}
-
fclose(proc_stat);
- return result;
+ if (!value_valid)
+ return (-1);
+
+ ps_submit_fork_rate (value.derive);
+ return (0);
}
+#endif /*KERNEL_LINUX */
-static void ps_submit_fork_rate (unsigned long value)
+#if KERNEL_SOLARIS
+static const char *ps_get_cmdline (pid_t pid, /* {{{ */
+ char *buffer, size_t buffer_size)
{
- value_t values[1];
- value_list_t vl = VALUE_LIST_INIT;
+ char path[PATH_MAX];
+ psinfo_t info;
+ int status;
- values[0].derive = (derive_t) value;
+ snprintf(path, sizeof (path), "/proc/%i/psinfo", pid);
- vl.values = values;
- vl.values_len = 1;
- sstrncpy (vl.host, hostname_g, sizeof (vl.host));
- sstrncpy (vl.plugin, "processes", sizeof (vl.plugin));
- sstrncpy (vl.plugin_instance, "", sizeof (vl.plugin_instance));
- sstrncpy (vl.type, "fork_rate", sizeof (vl.type));
- sstrncpy (vl.type_instance, "", sizeof (vl.type_instance));
+ status = read_file_contents (path, (void *) &info, sizeof (info));
+ if (status != ((int) buffer_size))
+ {
+ ERROR ("processes plugin: Unexpected return value "
+ "while reading \"%s\": "
+ "Returned %i but expected %zu.",
+ path, status, buffer_size);
+ return (NULL);
+ }
- plugin_dispatch_values (&vl);
+ info.pr_psargs[sizeof (info.pr_psargs) - 1] = 0;
+ sstrncpy (buffer, info.pr_psargs, buffer_size);
+
+ return (buffer);
+} /* }}} int ps_get_cmdline */
+
+/*
+ * Reads process information on the Solaris OS. The information comes mainly from
+ * /proc/PID/status, /proc/PID/psinfo and /proc/PID/usage
+ * The values for input and ouput chars are calculated "by hand"
+ * Added a few "solaris" specific process states as well
+ */
+static int ps_read_process(int pid, procstat_t *ps, char *state)
+{
+ char filename[64];
+ char f_psinfo[64], f_usage[64];
+ char *buffer;
+
+ pstatus_t *myStatus;
+ psinfo_t *myInfo;
+ prusage_t *myUsage;
+
+ snprintf(filename, sizeof (filename), "/proc/%i/status", pid);
+ snprintf(f_psinfo, sizeof (f_psinfo), "/proc/%i/psinfo", pid);
+ snprintf(f_usage, sizeof (f_usage), "/proc/%i/usage", pid);
+
+
+ buffer = malloc(sizeof (pstatus_t));
+ memset(buffer, 0, sizeof (pstatus_t));
+ read_file_contents(filename, buffer, sizeof (pstatus_t));
+ myStatus = (pstatus_t *) buffer;
+
+ buffer = malloc(sizeof (psinfo_t));
+ memset(buffer, 0, sizeof(psinfo_t));
+ read_file_contents(f_psinfo, buffer, sizeof (psinfo_t));
+ myInfo = (psinfo_t *) buffer;
+
+ buffer = malloc(sizeof (prusage_t));
+ memset(buffer, 0, sizeof(prusage_t));
+ read_file_contents(f_usage, buffer, sizeof (prusage_t));
+ myUsage = (prusage_t *) buffer;
+
+ sstrncpy(ps->name, myInfo->pr_fname, sizeof (myInfo->pr_fname));
+ ps->num_lwp = myStatus->pr_nlwp;
+ if (myInfo->pr_wstat != 0) {
+ ps->num_proc = 0;
+ ps->num_lwp = 0;
+ *state = (char) 'Z';
+ return (0);
+ } else {
+ ps->num_proc = 1;
+ ps->num_lwp = myInfo->pr_nlwp;
+ }
+
+ /*
+ * Convert system time and user time from nanoseconds to microseconds
+ * for compatibility with the linux module
+ */
+ ps->cpu_system_counter = myStatus -> pr_stime.tv_nsec / 1000;
+ ps->cpu_user_counter = myStatus -> pr_utime.tv_nsec / 1000;
+
+ /*
+ * Convert rssize from KB to bytes to be consistent w/ the linux module
+ */
+ ps->vmem_rss = myInfo->pr_rssize * 1024;
+ ps->vmem_size = myInfo->pr_size * 1024;
+ ps->vmem_minflt_counter = myUsage->pr_minf;
+ ps->vmem_majflt_counter = myUsage->pr_majf;
+
+ /*
+ * TODO: Data and code segment calculations for Solaris
+ */
+
+ ps->vmem_data = -1;
+ ps->vmem_code = -1;
+ ps->stack_size = myStatus->pr_stksize;
+
+ /*
+ * Calculating input/ouput chars
+ * Formula used is total chars / total blocks => chars/block
+ * then convert input/output blocks to chars
+ */
+ ulong_t tot_chars = myUsage->pr_ioch;
+ ulong_t tot_blocks = myUsage->pr_inblk + myUsage->pr_oublk;
+ ulong_t chars_per_block = 1;
+ if (tot_blocks != 0)
+ chars_per_block = tot_chars / tot_blocks;
+ ps->io_rchar = myUsage->pr_inblk * chars_per_block;
+ ps->io_wchar = myUsage->pr_oublk * chars_per_block;
+ ps->io_syscr = myUsage->pr_sysc;
+ ps->io_syscw = myUsage->pr_sysc;
+
+
+ /*
+ * TODO: Find way of setting BLOCKED and PAGING status
+ */
+
+ *state = (char) 'R';
+ if (myStatus->pr_flags & PR_ASLEEP)
+ *state = (char) 'S';
+ else if (myStatus->pr_flags & PR_STOPPED)
+ *state = (char) 'T';
+ else if (myStatus->pr_flags & PR_DETACH)
+ *state = (char) 'E';
+ else if (myStatus->pr_flags & PR_DAEMON)
+ *state = (char) 'A';
+ else if (myStatus->pr_flags & PR_ISSYS)
+ *state = (char) 'Y';
+ else if (myStatus->pr_flags & PR_ORPHAN)
+ *state = (char) 'O';
+
+ sfree(myStatus);
+ sfree(myInfo);
+ sfree(myUsage);
+
+ return (0);
}
-#endif /* KERNEL_LINUX */
+/*
+ * Reads the number of threads created since the last reboot. On Solaris these
+ * are retrieved from kstat (module cpu, name sys, class misc, stat nthreads).
+ * The result is the sum for all the threads created on each cpu
+ */
+static int read_fork_rate()
+{
+ extern kstat_ctl_t *kc;
+ kstat_t *ksp_chain = NULL;
+ derive_t result = 0;
+
+ if (kc == NULL)
+ return (-1);
+
+ for (ksp_chain = kc->kc_chain;
+ ksp_chain != NULL;
+ ksp_chain = ksp_chain->ks_next)
+ {
+ if ((strcmp (ksp_chain->ks_module, "cpu") == 0)
+ && (strcmp (ksp_chain->ks_name, "sys") == 0)
+ && (strcmp (ksp_chain->ks_class, "misc") == 0))
+ {
+ long long tmp;
+
+ kstat_read (kc, ksp_chain, NULL);
+
+ tmp = get_kstat_value(ksp_chain, "nthreads");
+ if (tmp != -1LL)
+ result += tmp;
+ }
+ }
+
+ ps_submit_fork_rate (result);
+ return (0);
+}
+#endif /* KERNEL_SOLARIS */
#if HAVE_THREAD_INFO
static int mach_get_task_name (task_t t, int *pid, char *name, size_t name_max_len)
procstat_entry_t pse;
char state;
- unsigned long fork_rate;
-
procstat_t *ps_ptr;
running = sleeping = zombies = stopped = paging = blocked = 0;
for (ps_ptr = list_head_g; ps_ptr != NULL; ps_ptr = ps_ptr->next)
ps_submit_proc_list (ps_ptr);
- fork_rate = read_fork_rate();
- if (fork_rate != ULONG_MAX)
- ps_submit_fork_rate(fork_rate);
+ read_fork_rate();
/* #endif KERNEL_LINUX */
#elif HAVE_LIBKVM_GETPROCS && HAVE_STRUCT_KINFO_PROC_FREEBSD
kvm_t *kd;
char errbuf[1024];
- char cmdline[ARG_MAX];
- char *cmdline_ptr;
- struct kinfo_proc *procs; /* array of processes */
- char **argv;
- int count; /* returns number of processes */
+ struct kinfo_proc *procs; /* array of processes */
+ struct kinfo_proc *proc_ptr = NULL;
+ int count; /* returns number of processes */
int i;
procstat_t *ps_ptr;
/* Iterate through the processes in kinfo_proc */
for (i = 0; i < count; i++)
{
- /* retrieve the arguments */
- cmdline[0] = 0;
- cmdline_ptr = NULL;
-
- argv = kvm_getargv (kd, (const struct kinfo_proc *) &(procs[i]), 0);
- if (argv != NULL)
+ /* Create only one process list entry per _process_, i.e.
+ * filter out threads (duplicate PID entries). */
+ if ((proc_ptr == NULL) || (proc_ptr->ki_pid != procs[i].ki_pid))
{
- int status;
- int argc;
+ char cmdline[ARG_MAX] = "";
+ _Bool have_cmdline = 0;
- argc = 0;
- while (argv[argc] != NULL)
- argc++;
-
- status = strjoin (cmdline, sizeof (cmdline),
- argv, argc, " ");
-
- if (status < 0)
- {
- WARNING ("processes plugin: Command line did "
- "not fit into buffer.");
- }
- else
+ proc_ptr = &(procs[i]);
+ /* Don't probe system processes and processes without arguments */
+ if (((procs[i].ki_flag & P_SYSTEM) == 0)
+ && (procs[i].ki_args != NULL))
{
- cmdline_ptr = &cmdline[0];
- }
- }
+ char **argv;
+ int argc;
+ int status;
+
+ /* retrieve the arguments */
+ argv = kvm_getargv (kd, proc_ptr, /* nchr = */ 0);
+ argc = 0;
+ if ((argv != NULL) && (argv[0] != NULL))
+ {
+ while (argv[argc] != NULL)
+ argc++;
+
+ status = strjoin (cmdline, sizeof (cmdline), argv, argc, " ");
+ if (status < 0)
+ WARNING ("processes plugin: Command line did not fit into buffer.");
+ else
+ have_cmdline = 1;
+ }
+ } /* if (process has argument list) */
- pse.id = procs[i].ki_pid;
- pse.age = 0;
+ pse.id = procs[i].ki_pid;
+ pse.age = 0;
- pse.num_proc = 1;
- pse.num_lwp = procs[i].ki_numthreads;
+ pse.num_proc = 1;
+ pse.num_lwp = procs[i].ki_numthreads;
- pse.vmem_size = procs[i].ki_size;
- pse.vmem_rss = procs[i].ki_rssize * getpagesize();
- pse.vmem_data = procs[i].ki_dsize * getpagesize();
- pse.vmem_code = procs[i].ki_tsize * getpagesize();
- pse.stack_size = procs[i].ki_ssize * getpagesize();
- pse.vmem_minflt = 0;
- pse.vmem_minflt_counter = procs[i].ki_rusage.ru_minflt;
- pse.vmem_majflt = 0;
- pse.vmem_majflt_counter = procs[i].ki_rusage.ru_majflt;
+ pse.vmem_size = procs[i].ki_size;
+ pse.vmem_rss = procs[i].ki_rssize * pagesize;
+ pse.vmem_data = procs[i].ki_dsize * pagesize;
+ pse.vmem_code = procs[i].ki_tsize * pagesize;
+ pse.stack_size = procs[i].ki_ssize * pagesize;
+ pse.vmem_minflt = 0;
+ pse.vmem_minflt_counter = procs[i].ki_rusage.ru_minflt;
+ pse.vmem_majflt = 0;
+ pse.vmem_majflt_counter = procs[i].ki_rusage.ru_majflt;
- pse.cpu_user = 0;
- pse.cpu_user_counter = procs[i].ki_rusage.ru_utime.tv_sec
- * 1000
- + procs[i].ki_rusage.ru_utime.tv_usec;
- pse.cpu_system = 0;
- pse.cpu_system_counter = procs[i].ki_rusage.ru_stime.tv_sec
- * 1000
- + procs[i].ki_rusage.ru_stime.tv_usec;
+ pse.cpu_user = 0;
+ pse.cpu_system = 0;
+ pse.cpu_user_counter = 0;
+ pse.cpu_system_counter = 0;
+ /*
+ * The u-area might be swapped out, and we can't get
+ * at it because we have a crashdump and no swap.
+ * If it's here fill in these fields, otherwise, just
+ * leave them 0.
+ */
+ if (procs[i].ki_flag & P_INMEM)
+ {
+ pse.cpu_user_counter = procs[i].ki_rusage.ru_utime.tv_usec
+ + (1000000lu * procs[i].ki_rusage.ru_utime.tv_sec);
+ pse.cpu_system_counter = procs[i].ki_rusage.ru_stime.tv_usec
+ + (1000000lu * procs[i].ki_rusage.ru_stime.tv_sec);
+ }
- /* no io data */
- pse.io_rchar = -1;
- pse.io_wchar = -1;
- pse.io_syscr = -1;
- pse.io_syscw = -1;
+ /* no I/O data */
+ pse.io_rchar = -1;
+ pse.io_wchar = -1;
+ pse.io_syscr = -1;
+ pse.io_syscw = -1;
+
+ ps_list_add (procs[i].ki_comm, have_cmdline ? cmdline : NULL, &pse);
+ } /* if ((proc_ptr == NULL) || (proc_ptr->ki_pid != procs[i].ki_pid)) */
switch (procs[i].ki_stat)
{
case SLOCK: blocked++; break;
case SZOMB: zombies++; break;
}
-
- ps_list_add (procs[i].ki_comm, cmdline_ptr, &pse);
}
kvm_close(kd);
if (procentry[i].pi_pid == 0)
cmdline = "swapper";
cargs = cmdline;
- }
+ }
else
{
if (getargs(&procentry[i], sizeof(struct procentry64), arglist, MAXARGLN) >= 0)
for (ps = list_head_g; ps != NULL; ps = ps->next)
ps_submit_proc_list (ps);
-#endif /* HAVE_PROCINFO_H */
+/* #endif HAVE_PROCINFO_H */
+
+#elif KERNEL_SOLARIS
+ /*
+ * The Solaris section adds a few more process states and removes some
+ * process states compared to linux. Most notably there is no "PAGING"
+ * and "BLOCKED" state for a process. The rest is similar to the linux
+ * code.
+ */
+ int running = 0;
+ int sleeping = 0;
+ int zombies = 0;
+ int stopped = 0;
+ int detached = 0;
+ int daemon = 0;
+ int system = 0;
+ int orphan = 0;
+
+ struct dirent *ent;
+ DIR *proc;
+
+ int status;
+ procstat_t *ps_ptr;
+ char state;
+
+ char cmdline[PRARGSZ];
+
+ ps_list_reset ();
+
+ proc = opendir ("/proc");
+ if (proc == NULL)
+ return (-1);
+
+ while ((ent = readdir(proc)) != NULL)
+ {
+ int pid;
+ struct procstat ps;
+ procstat_entry_t pse;
+
+ if (!isdigit ((int) ent->d_name[0]))
+ continue;
+
+ if ((pid = atoi (ent->d_name)) < 1)
+ continue;
+
+ status = ps_read_process (pid, &ps, &state);
+ if (status != 0)
+ {
+ DEBUG("ps_read_process failed: %i", status);
+ continue;
+ }
+
+ pse.id = pid;
+ pse.age = 0;
+
+ pse.num_proc = ps.num_proc;
+ pse.num_lwp = ps.num_lwp;
+ pse.vmem_size = ps.vmem_size;
+ pse.vmem_rss = ps.vmem_rss;
+ pse.vmem_data = ps.vmem_data;
+ pse.vmem_code = ps.vmem_code;
+ pse.stack_size = ps.stack_size;
+
+ pse.vmem_minflt = 0;
+ pse.vmem_minflt_counter = ps.vmem_minflt_counter;
+ pse.vmem_majflt = 0;
+ pse.vmem_majflt_counter = ps.vmem_majflt_counter;
+
+ pse.cpu_user = 0;
+ pse.cpu_user_counter = ps.cpu_user_counter;
+ pse.cpu_system = 0;
+ pse.cpu_system_counter = ps.cpu_system_counter;
+
+ pse.io_rchar = ps.io_rchar;
+ pse.io_wchar = ps.io_wchar;
+ pse.io_syscr = ps.io_syscr;
+ pse.io_syscw = ps.io_syscw;
+
+ switch (state)
+ {
+ case 'R': running++; break;
+ case 'S': sleeping++; break;
+ case 'E': detached++; break;
+ case 'Z': zombies++; break;
+ case 'T': stopped++; break;
+ case 'A': daemon++; break;
+ case 'Y': system++; break;
+ case 'O': orphan++; break;
+ }
+
+
+ ps_list_add (ps.name,
+ ps_get_cmdline ((pid_t) pid,
+ cmdline, sizeof (cmdline)),
+ &pse);
+ } /* while(readdir) */
+ closedir (proc);
+
+ ps_submit_state ("running", running);
+ ps_submit_state ("sleeping", sleeping);
+ ps_submit_state ("zombies", zombies);
+ ps_submit_state ("stopped", stopped);
+ ps_submit_state ("detached", detached);
+ ps_submit_state ("daemon", daemon);
+ ps_submit_state ("system", system);
+ ps_submit_state ("orphan", orphan);
+
+ for (ps_ptr = list_head_g; ps_ptr != NULL; ps_ptr = ps_ptr->next)
+ ps_submit_proc_list (ps_ptr);
+
+ read_fork_rate();
+#endif /* KERNEL_SOLARIS */
return (0);
} /* int ps_read */