remove self-defense options from config
This commit is contained in:
Alexey Avramov
parent
3b0644eac3
commit
a154ca038b
@ -200,6 +200,7 @@ Please create [issues](https://github.com/hakavlad/nohang/issues). Use cases, fe
|
||||
- Add `oom-trigger`
|
||||
- Adoption of the [code of conduct](https://github.com/hakavlad/nohang/blob/master/CODE_OF_CONDUCT.md)
|
||||
- Redesign of the config
|
||||
- Remove self-defense options from config, use systemd unit scheduling instead
|
||||
|
||||
- [v0.1](https://github.com/hakavlad/nohang/releases/tag/v0.1), 2018-11-23
|
||||
- 1st release
|
||||
|
||||
181
nohang
181
nohang
@ -2,6 +2,8 @@
|
||||
"""A daemon that prevents OOM in Linux systems."""
|
||||
from time import sleep, time
|
||||
|
||||
start_time = time()
|
||||
|
||||
import os
|
||||
from operator import itemgetter
|
||||
|
||||
@ -11,18 +13,9 @@ from argparse import ArgumentParser
|
||||
from sys import stdout
|
||||
from signal import SIGKILL, SIGTERM
|
||||
|
||||
import logging
|
||||
|
||||
start_time = time()
|
||||
|
||||
sig_dict = {SIGKILL: 'SIGKILL',
|
||||
SIGTERM: 'SIGTERM'}
|
||||
|
||||
'''
|
||||
nm = 30
|
||||
nc = nm + 1
|
||||
'''
|
||||
|
||||
self_uid = os.geteuid()
|
||||
self_pid = str(os.getpid())
|
||||
|
||||
@ -30,22 +23,9 @@ wait_time = 2
|
||||
cache_time = 30
|
||||
cache_path = '/dev/shm/nohang_env_cache'
|
||||
|
||||
|
||||
psi_path = '/proc/pressure/memory'
|
||||
|
||||
psi_support = os.path.exists(psi_path)
|
||||
|
||||
# Log configuration
|
||||
# TODO make it configurable from a config file?
|
||||
|
||||
logging.basicConfig(
|
||||
level=logging.INFO,
|
||||
format="\n[%(asctime)s] %(levelname)s\n%(message)s",
|
||||
)
|
||||
|
||||
logger = logging.getLogger('nohang')
|
||||
logger.setLevel(logging.DEBUG)
|
||||
|
||||
##########################################################################
|
||||
|
||||
|
||||
@ -667,8 +647,8 @@ def find_victim_and_send_signal(signal):
|
||||
new_value = stat_dict[key] + 1
|
||||
stat_dict.update({key: new_value})
|
||||
|
||||
logger.info(mem_info)
|
||||
logger.info(etc_info)
|
||||
print(mem_info)
|
||||
print(etc_info)
|
||||
|
||||
if gui_notifications:
|
||||
send_notify_etc(pid, name, command)
|
||||
@ -707,8 +687,8 @@ def find_victim_and_send_signal(signal):
|
||||
'Sending \033[4m{}\033[0m to the victim; {}'.format(
|
||||
victim_info, sig_dict[signal], send_result)
|
||||
|
||||
logger.info(mem_info)
|
||||
logger.info(preventing_oom_message)
|
||||
print(mem_info)
|
||||
print(preventing_oom_message)
|
||||
|
||||
stats_msg = '\033[4mUptime: {}; corrective actions:\033[0m'.format(
|
||||
format_time(time() - start_time))
|
||||
@ -716,7 +696,7 @@ def find_victim_and_send_signal(signal):
|
||||
for key in stat_dict:
|
||||
stats_msg += '\n - {}: {}'.format(key, stat_dict[key])
|
||||
|
||||
logger.info(stats_msg)
|
||||
print(stats_msg)
|
||||
|
||||
else:
|
||||
|
||||
@ -1016,8 +996,6 @@ except IndexError:
|
||||
print_config = conf_parse_bool('print_config')
|
||||
print_mem_check_results = conf_parse_bool('print_mem_check_results')
|
||||
print_sleep_periods = conf_parse_bool('print_sleep_periods')
|
||||
realtime_ionice = conf_parse_bool('realtime_ionice')
|
||||
mlockall = conf_parse_bool('mlockall')
|
||||
gui_low_memory_warnings = conf_parse_bool('gui_low_memory_warnings')
|
||||
gui_notifications = conf_parse_bool('gui_notifications')
|
||||
decrease_oom_score_adj = conf_parse_bool('decrease_oom_score_adj')
|
||||
@ -1053,51 +1031,6 @@ zram_max_warnings_kb, zram_max_warnings_mb, zram_max_warnings_percent = calculat
|
||||
'zram_max_warnings')
|
||||
|
||||
|
||||
if 'realtime_ionice_classdata' in config_dict:
|
||||
realtime_ionice_classdata = string_to_int_convert_test(
|
||||
config_dict['realtime_ionice_classdata'])
|
||||
if realtime_ionice_classdata is None:
|
||||
print('Invalid value of the "realtime_ionice_classdata" parameter.')
|
||||
print('Valid values are integers from the range [0; 7].')
|
||||
print('Exit')
|
||||
exit()
|
||||
if realtime_ionice_classdata < 0 or realtime_ionice_classdata > 7:
|
||||
print('Invalid value of the "realtime_ionice_classdata" parameter.')
|
||||
print('Valid values are integers from the range [0; 7].')
|
||||
print('Exit')
|
||||
exit()
|
||||
else:
|
||||
print('All the necessary parameters must be in the config')
|
||||
print('There is no "realtime_ionice_classdata" parameter in the config')
|
||||
exit()
|
||||
|
||||
|
||||
if 'niceness' in config_dict:
|
||||
niceness = string_to_int_convert_test(config_dict['niceness'])
|
||||
if niceness is None:
|
||||
print('Invalid niceness value, not integer\nExit')
|
||||
exit()
|
||||
if niceness < -20 or niceness > 19:
|
||||
print('niceness out of range [-20; 19]\nExit')
|
||||
exit()
|
||||
else:
|
||||
print('niceness not in config\nExit')
|
||||
exit()
|
||||
|
||||
|
||||
if 'oom_score_adj' in config_dict:
|
||||
oom_score_adj = string_to_int_convert_test(
|
||||
config_dict['oom_score_adj'])
|
||||
if oom_score_adj is None:
|
||||
print('Invalid oom_score_adj value, not integer\nExit')
|
||||
exit()
|
||||
if oom_score_adj < -1000 or oom_score_adj > 1000:
|
||||
print('oom_score_adj out of range [-1000; 1000]\nExit')
|
||||
exit()
|
||||
else:
|
||||
print('oom_score_adj not in config\nExit')
|
||||
exit()
|
||||
|
||||
|
||||
if 'rate_mem' in config_dict:
|
||||
rate_mem = string_to_float_convert_test(config_dict['rate_mem'])
|
||||
@ -1323,56 +1256,6 @@ else:
|
||||
swap_min_warnings_kb = swap_min_warnings_swap
|
||||
|
||||
|
||||
##########################################################################
|
||||
|
||||
# self-defense
|
||||
|
||||
|
||||
# возожно стоит убрать поддержку mlockall и ionice
|
||||
|
||||
# Increase priority
|
||||
try:
|
||||
os.nice(niceness)
|
||||
niceness_result = 'OK'
|
||||
except PermissionError:
|
||||
niceness_result = 'Fail'
|
||||
pass
|
||||
|
||||
# Deny self-killing
|
||||
try:
|
||||
with open('/proc/self/oom_score_adj', 'w') as file:
|
||||
file.write('{}\n'.format(oom_score_adj))
|
||||
oom_score_adj_result = 'OK'
|
||||
except PermissionError:
|
||||
oom_score_adj_result = 'Fail'
|
||||
except OSError:
|
||||
oom_score_adj_result = 'Fail'
|
||||
|
||||
# Deny process swapping
|
||||
if mlockall:
|
||||
from ctypes import CDLL
|
||||
result = CDLL('libc.so.6', use_errno=True).mlockall(3)
|
||||
if result is 0:
|
||||
mla_res = 'OK'
|
||||
else:
|
||||
mla_res = 'Fail'
|
||||
else:
|
||||
mla_res = ''
|
||||
|
||||
|
||||
if self_uid == 0:
|
||||
root = True
|
||||
decrease_res = 'OK'
|
||||
else:
|
||||
root = False
|
||||
decrease_res = 'Impossible'
|
||||
|
||||
|
||||
if root and realtime_ionice:
|
||||
os.system('ionice -c 1 -n {} -p {}'.format(
|
||||
realtime_ionice_classdata, self_pid))
|
||||
|
||||
|
||||
##########################################################################
|
||||
|
||||
if print_config:
|
||||
@ -1458,23 +1341,7 @@ if print_config:
|
||||
print('zram_max_warnings: {} MiB, {} %'.format(
|
||||
round(zram_max_warnings_mb), round(zram_max_warnings_percent, 1)))
|
||||
|
||||
print(
|
||||
'\n7. Preventing the slowing down of the program'
|
||||
'\n[displaying these options need fix]\n')
|
||||
print('mlockall: {} ({})'.format(mlockall, mla_res))
|
||||
print('niceness: {} ({})'.format(
|
||||
niceness, niceness_result
|
||||
))
|
||||
print('oom_score_adj: {} ({})'.format(
|
||||
oom_score_adj, oom_score_adj_result
|
||||
))
|
||||
|
||||
print('realtime_ionice: {} ({})'.format(realtime_ionice, ''))
|
||||
|
||||
if realtime_ionice:
|
||||
print('realtime_ionice_classdata: {}'.format(realtime_ionice_classdata))
|
||||
|
||||
print('\n8. Output verbosity\n')
|
||||
print('\n7. Output verbosity\n')
|
||||
print('print_config: {}'.format(print_config))
|
||||
print('print_mem_check_results: {}'.format(print_mem_check_results))
|
||||
print('print_sleep_periods: {}\n'.format(print_sleep_periods))
|
||||
@ -1582,7 +1449,6 @@ if psi_support and not ignore_psi:
|
||||
|
||||
avg_value = ''
|
||||
|
||||
|
||||
while True:
|
||||
|
||||
if psi_support and not ignore_psi:
|
||||
@ -1606,37 +1472,6 @@ while True:
|
||||
# print('PSI is OK or psi_min_sleep_time_after_action did not pass')
|
||||
pass
|
||||
|
||||
'''
|
||||
if psi_support:
|
||||
|
||||
ta1= time()
|
||||
dt = ta1 - ta0
|
||||
|
||||
if dt >= avg_min_time:
|
||||
|
||||
a1 = psi_mem_some_avg_total()
|
||||
avg = (a1 - a0) / (ta1 - ta0) / 10000
|
||||
|
||||
print(rline1(psi_path))
|
||||
print('PSI mem some avg: {}, PSI avg time: {}'.format(round(avg, 2), round(dt, 1)))
|
||||
ta0 = ta1
|
||||
a0 = a1
|
||||
|
||||
|
||||
if avg >= sigkill_psi and time() - kill_psi_t0 >= psi_min_sleep_time_after_action:
|
||||
time0 = time()
|
||||
mem_info = 'avg ({}) > sigkill_psi ({})'.format(round(avg, 2), sigkill_psi)
|
||||
find_victim_and_send_signal(SIGKILL)
|
||||
kill_psi_t0 = time()
|
||||
elif avg >= sigterm_psi and time() - term_psi_t0 >= psi_min_sleep_time_after_action:
|
||||
time0 = time()
|
||||
mem_info = 'avg ({}) > sigterm_psi ({})'.format(round(avg, 2), sigterm_psi)
|
||||
find_victim_and_send_signal(SIGTERM)
|
||||
term_psi_t0 = time()
|
||||
else:
|
||||
print('PSI is OK or psi_min_sleep_time_after_action did not pass')
|
||||
'''
|
||||
|
||||
mem_available, swap_total, swap_free = check_mem_and_swap()
|
||||
|
||||
# if swap_min_sigkill is set in percent
|
||||
|
||||
70
nohang.conf
70
nohang.conf
@ -25,26 +25,11 @@
|
||||
6. GUI notifications:
|
||||
- OOM prevention results and
|
||||
- low memory warnings
|
||||
7. Preventing the slowing down of the program
|
||||
8. Output verbosity
|
||||
7. Output verbosity
|
||||
|
||||
Just read the description of the parameters and edit the values.
|
||||
Please restart the program after editing the config.
|
||||
|
||||
#####################################################################
|
||||
|
||||
Not imlemented.
|
||||
|
||||
$IGNORE_SWAPSPACE = FALSE
|
||||
|
||||
$IGNORE_ZRAM = TRUE
|
||||
|
||||
$IGNORE_PSI = TRUE
|
||||
$SIGKILL_PSI_AVG = 80
|
||||
$SIGTERM_PSI_AVG = 60
|
||||
$PSI_AVG_TIME = 3
|
||||
|
||||
|
||||
#####################################################################
|
||||
|
||||
1. Thresholds below which a signal should be sent to the victim
|
||||
@ -296,58 +281,7 @@ zram_max_warnings = 40 %
|
||||
|
||||
#####################################################################
|
||||
|
||||
|
||||
This settings broken!
|
||||
|
||||
7. Preventing the slowing down of the program
|
||||
|
||||
mlockall() lock ... all of the calling process's virtual address
|
||||
space into RAM, preventing that memory from being paged to the
|
||||
swap area. - `man mlockall`
|
||||
|
||||
It is disabled by default because the value mlockall = True in
|
||||
Fedora 28 causes the process to increase memory consumption by
|
||||
200 MiB. On Debian 8, Debian 9, Ubuntu 16.04 there is no such
|
||||
problem. Other distros is not tested.
|
||||
|
||||
mlockall = False
|
||||
|
||||
Установка отрицательных значений niceness и oom_score_adj
|
||||
требует наличия root прав.
|
||||
|
||||
Установка отрицательного niceness повышает приоритет процесса.
|
||||
|
||||
Valid values are integers from the range [-20; 19].
|
||||
|
||||
niceness = 0
|
||||
|
||||
Возможно этот параметр можно убрать, потому что теперь запрет
|
||||
самоубийства включен по умолчанию: nohang исключает себя
|
||||
из поиска жертв.
|
||||
|
||||
Set oom_score_adj for the nohang process.
|
||||
Valid values are integers from the range [-1000; 1000].
|
||||
Setting the values to -1000 will prohibit suicide.
|
||||
|
||||
oom_score_adj = 0
|
||||
|
||||
Read `man ionice` to understand the following parameters.
|
||||
Setting the True value requires the root privileges.
|
||||
|
||||
Не замечено большой пользы от этой опции.
|
||||
|
||||
realtime_ionice = False
|
||||
|
||||
'For realtime and best-effort, 0-7 are valid data
|
||||
(priority levels), and 0 represents the highest priority level.'
|
||||
- `man ionice`
|
||||
Valid values are integers from the range [0; 7].
|
||||
|
||||
realtime_ionice_classdata = 4
|
||||
|
||||
#####################################################################
|
||||
|
||||
8. Output verbosity
|
||||
7. Output verbosity
|
||||
|
||||
Display the configuration when the program starts.
|
||||
Valid values are True and False.
|
||||
|
||||
Loading…
Reference in New Issue
Block a user