def main(argv):
if len(argv) < 4:
print("error: parameters missing")
return
cache = CacheProxy('plugin_process')
last_config_version = cache.get('version')
config_version = int(argv[2])
host_id = argv[3]
if last_config_version is None or config_version != last_config_version:
config_client = ConfigClient()
current_version, config_process_list = config_client.get_user_config(
'plugin_process', host_id)
if config_process_list is not None:
cache.set('process_list', config_process_list)
cache.set('version', current_version)
else:
config_process_list = cache.get('process_list')
if not config_process_list:
cache.close()
return
ntp_checked, timestamp = time_util.get_ntp_time()
cpu_total_jiffies = cache.counter_to_gauge('cpu_total_jiffies',
get_cpu_total_jiffies())
total_mem = get_total_mem()
pids = get_pids()
process_info_list = []
page_size = resource.getpagesize()
for pid in pids:
stat_path = '/proc/%d/stat' % pid
if not os.path.isfile(stat_path):
continue
try:
with open(stat_path, 'r') as f_stat:
line = f_stat.readline()
values = line.split(None)
if len(values) < 24:
continue
name = values[1][1:len(values[1]) - 1]
cmdline_path = '/proc/%d/cmdline' % pid
if os.path.isfile(cmdline_path) and os.access(
cmdline_path, os.R_OK):
with open(cmdline_path, 'r') as f_cmd:
cmdline = f_cmd.readline().replace('\0', ' ').strip()
if cmdline:
name = cmdline
for p in config_process_list:
if fnmatch(name, p):
process_info = {'pid': pid, 'name': name, 'match': p}
status = values[2]
ppid = values[3]
process_info['parent_pid'] = ppid
process_info['proc_stat_cd'] = status
used_cpu_jiff = cache.counter_to_gauge(
'used_cpu_jiff_%d' % pid,
long(values[13]) + long(values[14]))
if used_cpu_jiff is None or cpu_total_jiffies is None:
cpu_usert = 0.0
else:
cpu_usert = used_cpu_jiff * 100.0 / cpu_total_jiffies
mem = long(values[23]) * page_size
if total_mem is None:
mem_usert = 0.0
else:
mem_usert = mem * 100.0 / total_mem / 1024
vir_mem = float(values[22]) / 1024.0
thread_num = int(values[19])
process_info['cpu_usert'] = cpu_usert
process_info['mem_usert'] = mem_usert
process_info['mem_byt_cnt'] = vir_mem
process_info['thd_cnt'] = thread_num
process_info_list.append(process_info)
except Exception:
pass
out_list = []
for p in config_process_list:
pid_list = []
process_count = 0
tot_cpu_usert = 0.0
tot_mem_usert = 0.0
tot_mem_byt_cnt = 0.0
tot_thd_cnt = 0
for process_info in process_info_list:
if process_info['match'] == p:
process_count += 1
tot_cpu_usert += process_info['cpu_usert']
tot_mem_usert += process_info['mem_usert']
tot_mem_byt_cnt += process_info['mem_byt_cnt']
tot_thd_cnt += process_info['thd_cnt']
pid_list.append(process_info['pid'])
dimensions = {'proc_name': p}
pid_list_record = cache.get('pip_list_record_' + p)
cache.set('pip_list_record_' + p, pid_list)
if pid_list_record is None or len(pid_list_record) == 0:
if len(pid_list) > 0:
is_process_up = 1
else:
is_process_up = 0
else:
is_process_up = 1
for pid in pid_list_record:
if pid not in pid_list:
is_process_up = 0
break
if process_count == 0:
metrics = {
'is_process_up': is_process_up,
'process_count': process_count,
'avg_cpu_usert': 0.0,
'avg_mem_usert': 0.0,
'avg_mem_byt_cnt': 0.0,
'avg_thd_cnt': 0,
'tot_cpu_usert': tot_cpu_usert,
'tot_mem_usert': tot_mem_usert,
'tot_mem_byt_cnt': tot_mem_byt_cnt,
'tot_thd_cnt': tot_thd_cnt
}
else:
metrics = {
'is_process_up': is_process_up,
'process_count': process_count,
'avg_cpu_usert': tot_cpu_usert / process_count,
'avg_mem_usert': tot_mem_usert / process_count,
'avg_mem_byt_cnt': tot_mem_byt_cnt / process_count,
'avg_thd_cnt': tot_thd_cnt / float(process_count),
'tot_cpu_usert': tot_cpu_usert,
'tot_mem_usert': tot_mem_usert,
'tot_mem_byt_cnt': tot_mem_byt_cnt,
'tot_thd_cnt': tot_thd_cnt
}
out = {
'dimensions': dimensions,
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
print(json.dumps(out_list))
sys.stdout.flush()
cache.close()
def main(argv):
cache = CacheProxy('cpu')
f = open('/proc/stat')
out_list = []
avg_used = 0
max_used = 0.0
avg_idle = 0
avg_user = 0
avg_sys = 0
avg_nice = 0
avg_irq = 0
avg_softirq = 0
avg_io_wait = 0
count_used = 0
count_idle = 0
count_user = 0
count_sys = 0
count_nice = 0
count_irq = 0
count_softirq = 0
count_io_wait = 0
ntp_checked, timestamp = time_util.get_ntp_time()
for line in f:
values = line.split(None)
name_len = len(values[0])
if name_len > 3 and values[0][0:3] == 'cpu':
cpu_id = int(values[0][3:name_len])
dimensions = {'cpu_idx': cpu_id}
metrics = {}
user_cnt = long(values[1])
nice_cnt = long(values[2])
system_cnt = long(values[3])
idle_cnt = long(values[4])
iowait_cnt = long(values[5])
irq_cnt = long(values[6])
softirq_cnt = long(values[7])
total_cnt = user_cnt + nice_cnt + system_cnt + idle_cnt + iowait_cnt + irq_cnt + softirq_cnt
user = cache.counter_to_gauge('user_%d' % cpu_id, user_cnt)
nice = cache.counter_to_gauge('nice_%d' % cpu_id, nice_cnt)
system = cache.counter_to_gauge('system_%d' % cpu_id, system_cnt)
idle = cache.counter_to_gauge('idle_%d' % cpu_id, idle_cnt)
iowait = cache.counter_to_gauge('iowait_%d' % cpu_id, iowait_cnt)
irq = cache.counter_to_gauge('irq_%d' % cpu_id, irq_cnt)
softirq = cache.counter_to_gauge('softirq_%d' % cpu_id,
softirq_cnt)
total = cache.counter_to_gauge('total_%d' % cpu_id, total_cnt)
if total is None or total == 0:
continue
if user is not None:
metrics['user_rto'] = user * 100.0 / total
avg_user += metrics['user_rto']
count_user += 1
if system is not None:
metrics['sys_rto'] = system * 100.0 / total
avg_sys += metrics['sys_rto']
count_sys += 1
if nice is not None:
metrics['nice_rto'] = nice * 100.0 / total
avg_nice += metrics['nice_rto']
count_nice += 1
if idle is not None:
metrics['idle_rto'] = idle * 100.0 / total
avg_idle += metrics['idle_rto']
count_idle += 1
if irq is not None:
metrics['irq_rto'] = irq * 100.0 / total
avg_irq += metrics['irq_rto']
count_irq += 1
if softirq is not None:
metrics['softirq_rto'] = softirq * 100.0 / total
avg_softirq += metrics['softirq_rto']
count_softirq += 1
if iowait is not None:
metrics['io_wait_rto'] = iowait * 100.0 / total
avg_io_wait += metrics['io_wait_rto']
count_io_wait += 1
if user is not None and nice is not None \
and system is not None and iowait is not None \
and irq is not None and softirq is not None:
used = user + nice + system + iowait + irq + softirq
metrics['used_rto'] = used * 100.0 / total
avg_used += metrics['used_rto']
if metrics['used_rto'] > max_used:
max_used = metrics['used_rto']
count_used += 1
if metrics:
out = {
'dimensions': dimensions,
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
metrics = {}
if count_used > 0:
metrics['avg_cpu_used_rto'] = avg_used / count_used
metrics['max_cpu_used_rto'] = max_used
if count_idle > 0:
metrics['avg_cpu_idle_rto'] = avg_idle / count_idle
if count_user > 0:
metrics['avg_cpu_user_rto'] = avg_user / count_user
if count_sys > 0:
metrics['avg_cpu_sys_rto'] = avg_sys / count_sys
if count_nice > 0:
metrics['avg_nice_rto'] = avg_nice / count_nice
if count_irq > 0:
metrics['avg_irq_rto'] = avg_irq / count_irq
if count_softirq > 0:
metrics['avg_softirq_rto'] = avg_softirq / count_softirq
if count_io_wait > 0:
metrics['avg_io_wait_rto'] = avg_io_wait / count_io_wait
# if cpu_count > 0:
# metrics = {'avg_cpu_used_rto': avg_used / cpu_count, 'max_cpu_used_rto': max_used,
# 'avg_cpu_idle_rto': avg_idle / cpu_count, 'avg_cpu_user_rto': avg_user / cpu_count,
# 'avg_cpu_sys_rto': avg_sys / cpu_count, 'avg_nice_rto': avg_nice / cpu_count,
# 'avg_irq_rto': avg_irq / cpu_count, 'avg_softirq_rto': avg_softirq / cpu_count,
# 'avg_io_wait_rto': avg_io_wait / cpu_count}
if metrics:
out = {
'dimensions': {
'schema_type': 'svr'
},
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
if out_list:
print(json.dumps(out_list))
sys.stdout.flush()
f.close()
cache.close()
def main(argv):
cache = CacheProxy('disk')
f_diskstats = open("/proc/diskstats")
out_list = []
avg_read_byt = 0.0
avg_write_byt = 0.0
avg_read = 0.0
avg_write = 0.0
max_read_byt = 0
max_write_byt = 0
max_read = 0
max_write = 0
# disk_count = 0
count_read_byt = 0
count_write_byt = 0
count_read = 0
count_write = 0
last_disk_name = None
ntp_checked, timestamp = time_util.get_ntp_time()
for line in f_diskstats:
values = line.split(None)
if values[3] == "0":
continue
# disk_count += 1
disk_idx = values[2]
dimensions = {'disk_idx': disk_idx}
metrics = {}
rd_ios = cache.counter_to_gauge('rd_ios_%s' % disk_idx,
long(values[3]))
rd_sectors = cache.counter_to_gauge('rd_sectors_%s' % disk_idx,
long(values[5]))
# rd_ticks = cache.counter_to_gauge('rd_ticks_%s' % disk_idx, long(values[6]))
wr_ios = cache.counter_to_gauge('wr_ios_%s' % disk_idx,
long(values[7]))
wr_sectors = cache.counter_to_gauge('wr_sectors_%s' % disk_idx,
long(values[9]))
# wr_ticks = cache.counter_to_gauge('wr_ticks_%s' % disk_idx, long(values[10]))
io_ticks = cache.counter_to_gauge('io_ticks_%s' % disk_idx,
long(values[12]))
time_in_queue = cache.counter_to_gauge('time_in_queue_%s' % disk_idx,
long(values[13]))
if rd_ios is None or wr_ios is None:
continue
if last_disk_name and disk_idx.find(last_disk_name) > -1:
continue
else:
last_disk_name = disk_idx
read_byt_cnt = None
write_byt_cnt = None
await_tm = None
avgrq_sz = None
svc_tm = None
if rd_sectors is not None:
read_byt_cnt = rd_sectors * 512
if wr_sectors is not None:
write_byt_cnt = wr_sectors * 512
if rd_sectors is not None and wr_sectors is not None:
if (rd_ios + wr_ios) > 0:
avgrq_sz = (rd_sectors + wr_sectors) / float(rd_ios + wr_ios)
else:
avgrq_sz = 0.0
if time_in_queue is not None:
avgqu_sz = time_in_queue / 1000.0 / 60.0
if time_in_queue is not None:
if (rd_ios + wr_ios) > 0:
await_tm = time_in_queue / float(rd_ios + wr_ios)
else:
await_tm = 0.0
if io_ticks is not None:
if (rd_ios + wr_ios) > 0:
svc_tm = io_ticks / float(rd_ios + wr_ios)
else:
svc_tm = 0.0
if io_ticks is not None:
used_rto = io_ticks * 100 / 1000.0 / 60.0
if read_byt_cnt is not None:
metrics['read_byt_cnt'] = read_byt_cnt / 60.0
avg_read_byt += read_byt_cnt
if read_byt_cnt > max_read_byt:
max_read_byt = read_byt_cnt
count_read_byt += 1
if write_byt_cnt is not None:
metrics['write_byt_cnt'] = write_byt_cnt / 60.0
avg_write_byt += write_byt_cnt
if write_byt_cnt > max_write_byt:
max_write_byt = write_byt_cnt
count_write_byt += 1
if rd_ios is not None:
metrics['read_cnt'] = rd_ios / 60.0
avg_read += rd_ios
if rd_ios > max_read:
max_read = rd_ios
count_read += 1
if wr_ios is not None:
metrics['write_cnt'] = wr_ios / 60.0
avg_write += wr_ios
if wr_ios > max_write:
max_write = wr_ios
count_write += 1
if avgrq_sz is not None:
metrics['avgrq_sz'] = avgrq_sz
if avgqu_sz is not None:
metrics['avgqu_sz'] = avgqu_sz
if await_tm is not None:
metrics['await_tm'] = await_tm
if svc_tm is not None:
metrics['svc_tm'] = svc_tm
if used_rto is not None:
metrics['used_rto'] = used_rto
if metrics:
out = {
'dimensions': dimensions,
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
# if disk_count > 0:
# metrics = {}
#
# metrics['avg_read_byt_cnt'] = avg_read_byt / 60.0 / disk_count
# metrics['avg_write_byt_cnt'] = avg_write_byt / 60.0 / disk_count
# metrics['avg_read_cnt'] = avg_read / 60.0 / disk_count
# metrics['avg_write_cnt'] = avg_write / 60.0 / disk_count
# metrics['max_read_byt_cnt'] = max_read_byt
# metrics['max_write_byt_cnt'] = max_write_byt
# metrics['max_read_cnt'] = max_read
# metrics['max_write_cnt'] = max_write
#
# out = {'dimensions': {'schema_type': 'svr'},
# 'metrics': metrics,
# 'timestamp': timestamp,
# 'ntp_checked': ntp_checked}
# out_list.append(out)
metrics = {}
if count_read_byt > 0:
metrics['avg_read_byt_cnt'] = avg_read_byt / 60.0 / count_read_byt
metrics['max_read_byt_cnt'] = max_read_byt
if count_write_byt > 0:
metrics['avg_write_byt_cnt'] = avg_write_byt / 60.0 / count_write_byt
metrics['max_write_byt_cnt'] = max_write_byt
if count_read > 0:
metrics['avg_read_cnt'] = avg_read / 60.0 / count_read
metrics['max_read_cnt'] = max_read
if count_write > 0:
metrics['avg_write_cnt'] = avg_write / 60.0 / count_write
metrics['max_write_cnt'] = max_write
if metrics:
out = {
'dimensions': {
'schema_type': 'svr'
},
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
if out_list:
print(json.dumps(out_list))
sys.stdout.flush()
f_diskstats.close()
cache.close()
def main(argv):
HZ = get_HZ()
# print HZ
total_mem = get_total_mem()
# print total_mem
cache = CacheProxy('process')
cpu_total_jiffies = cache.counter_to_gauge('cpu_total_jiffies',
get_cpu_total_jiffies())
pids = get_pids()
out_list = []
mem_usert_total = 0.0
count = 0
top10 = []
ntp_checked, timestamp = time_util.get_ntp_time()
page_size = resource.getpagesize()
for pid in pids:
stat_path = '/proc/%d/stat' % pid
if not os.path.isfile(stat_path):
continue
try:
with open(stat_path, 'r') as f_stat:
line = f_stat.readline()
values = line.split(None)
if len(values) < 24:
continue
name = values[1][1:len(values[1]) - 1]
status = values[2]
ppid = values[3]
dimensions = {
'proc_id': pid,
'parent_pid': ppid,
'proc_nm': name,
'proc_stat_cd': status
}
metrics = {}
used_cpu_jiff = cache.counter_to_gauge(
'used_cpu_jiff_%d' % pid,
long(values[13]) + long(values[14]))
if used_cpu_jiff is None or cpu_total_jiffies is None:
cpu_usert = None
t_cpu_usert = None
else:
cpu_usert = used_cpu_jiff * 100.0 / cpu_total_jiffies
t_cpu_usert = cpu_usert
mem = long(values[23]) * page_size
if total_mem is None:
mem_usert = 0.0
else:
mem_usert = mem * 100.0 / total_mem / 1024
vir_mem = float(values[22]) / 1024.0
cpu_time = (float(values[13]) + float(values[14])) / HZ
priority = int(values[17])
nice = int(values[18])
thread_num = int(values[19])
cpu_core_cnt = get_cpu_core_count()
time_now = time.time()
start_time = time_now - get_uptime() + float(values[21]) / HZ
# start_time_local = time.localtime(start_time)
# start_time_str = time.strftime("%Y-%m-%d %H:%M:%S", start_time_local)
dimensions['strt_ymdt'] = long(start_time * 1000)
dimensions['proc_cpu_cnt'] = cpu_core_cnt
if cpu_usert is not None:
metrics['cpu_usert'] = cpu_usert
if t_cpu_usert is not None:
metrics['proc_t_cpu_usert'] = t_cpu_usert
if mem_usert is not None:
metrics['p_proc_mem_usert'] = mem_usert
mem_usert_total += mem_usert
metrics['vir_mem_byt_cnt'] = vir_mem
metrics['cpu_tm_ss'] = cpu_time
metrics['prit_rnk'] = priority
metrics['nice_val'] = nice
metrics['thd_cnt'] = thread_num
if metrics:
out = {
'dimensions': dimensions,
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
top10_check_insert(top10, out)
# out_list.append(out)
count += 1
except Exception:
pass
for item in top10:
out_list.append(item)
out = {
'dimensions': {
'schema_type': 'svr'
},
'metrics': {
'proc_cnt': count,
'proc_mem_usert': mem_usert_total,
'run_que_len': get_run_que_len()
},
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list.append(out)
print(json.dumps(out_list))
sys.stdout.flush()
cache.close()
def main(argv):
cache = CacheProxy('memory')
f_mem_info = open('/proc/meminfo')
for line in f_mem_info:
values = line.split(None)
if values[0] == 'MemTotal:':
mem_mb = long(values[1]) / 1024.0
if values[0] == 'MemFree:':
free_mem_mb = long(values[1]) / 1024.0
# used_mem_mb = mem_mb - free_mem_mb
# mem_usert = used_mem_mb * 100 / mem_mb
if values[0] == 'Shmem:':
shared_mem_mb = long(values[1]) / 1024.0
if values[0] == 'Buffers:':
bffr_mb = long(values[1]) / 1024.0
if values[0] == 'Cached:':
cache_mb = long(values[1]) / 1024.0
if values[0] == 'SwapTotal:':
swap_mb = long(values[1]) / 1024.0
if values[0] == 'SwapFree:':
if swap_mb > 0.0:
swap_free_mb = long(values[1]) / 1024.0
swap_used_mb = swap_mb - swap_free_mb
swap_usert = swap_used_mb * 100 / swap_mb
else:
swap_free_mb = 0.0
swap_used_mb = 0.0
swap_usert = 0.0
used_mem_mb = mem_mb - free_mem_mb - bffr_mb - cache_mb
mem_usert = used_mem_mb * 100 / mem_mb
f_mem_info.close()
f_vmstat = open('/proc/vmstat')
for line in f_vmstat:
values = line.split(None)
if values[0] == 'pgpgin':
pgin_mb = cache.counter_to_gauge('pgin', long(values[1]))
if values[0] == 'pgpgout':
pgout_mb = cache.counter_to_gauge('pgout', long(values[1]))
if values[0] == 'pgfault':
pgfault = cache.counter_to_gauge('pgfault', long(values[1]))
f_vmstat.close()
cache.close()
dimensions = {}
metrics = {}
ntp_checked, timestamp = time_util.get_ntp_time()
metrics['mem_mb'] = mem_mb
metrics['free_mem_mb'] = free_mem_mb
metrics['used_mem_mb'] = used_mem_mb
metrics['mem_usert'] = mem_usert
if 'shared_mem_mb' in vars():
metrics['shared_mem_mb'] = shared_mem_mb
metrics['bffr_mb'] = bffr_mb
metrics['cache_mb'] = cache_mb
metrics['swap_mb'] = swap_mb
metrics['swap_free_mb'] = swap_free_mb
metrics['swap_used_mb'] = swap_used_mb
metrics['swap_usert'] = swap_usert
if pgin_mb is not None:
metrics['pgin_mb'] = pgin_mb / 1024.0 / 60
if pgout_mb is not None:
metrics['pgout_mb'] = pgout_mb / 1024.0 / 60
if pgfault is not None:
metrics['pgfault_tcnt'] = pgfault / 60
out = {
'dimensions': dimensions,
'metrics': metrics,
'timestamp': timestamp,
'ntp_checked': ntp_checked
}
out_list = [out]
print(json.dumps(out_list))
sys.stdout.flush()