# -*- coding: Latin-1 -*- # # Linux OOM Analyser # # Copyright (c) 2017-2020 Carsten Grohmann # License: MIT - THIS PROGRAM COMES WITH NO WARRANTY import re DEBUG = False """Show additional information during the development cycle""" VERSION = "0.4.0 (devel)" """Version number""" # __pragma__ ('skip') js_undefined = 0 # Prevent complaints by optional static checker # __pragma__ ('noskip') class OOMEntityState(object): """Simple enum to track the completeness of an OOM block""" empty = 1 invalid = 2 started = 3 complete = 4 def is_visible(element): return element.offsetWidth > 0 and element.offsetHeight > 0 def hide_element(element_id): """Hide the given HTML element""" element = document.getElementById(element_id) element.classList.add('js-text--display-none') def show_element(element_id): """Show the given HTML element""" element = document.getElementById(element_id) element.classList.remove('js-text--display-none') def toggle(element_id): """Toggle the visibility of the given HTML element""" element = document.getElementById(element_id) element.classList.toggle('js-text--display-none') def error(msg): """Show the error box and add the error message""" show_notifybox('ERROR', msg) def internal_error(msg): """Show the error box and add the internal error message""" show_notifybox('INTERNAL ERROR', msg) def warning(msg): """Show the error box and add the warning message""" show_notifybox('WARNING', msg) def show_notifybox(prefix, msg): """Show the error box and the message""" if prefix == 'WARNING': css_class = 'js-notify_box__msg--warning' else: css_class = 'js-notify_box__msg--error' show_element('notify_box') notify_box = document.getElementById('notify_box') notification = document.createElement('div') notification.classList.add(css_class) notification.innerHTML = '{}: {}
'.format(prefix, msg) notify_box.appendChild(notification) class KernelConfig(object): """Kernel configuration""" GFP_FLAGS = { 'GFP_ATOMIC': {'value': '__GFP_HIGH | __GFP_ATOMIC | __GFP_KSWAPD_RECLAIM'}, 'GFP_KERNEL': {'value': '__GFP_RECLAIM | __GFP_IO | __GFP_FS'}, 'GFP_KERNEL_ACCOUNT': {'value': 'GFP_KERNEL | __GFP_ACCOUNT'}, 'GFP_NOWAIT': {'value': '__GFP_KSWAPD_RECLAIM'}, 'GFP_NOIO': {'value': '__GFP_RECLAIM'}, 'GFP_NOFS': {'value': '__GFP_RECLAIM | __GFP_IO'}, 'GFP_USER': {'value': '__GFP_RECLAIM | __GFP_IO | __GFP_FS | __GFP_HARDWALL'}, 'GFP_DMA': {'value': '__GFP_DMA'}, 'GFP_DMA32': {'value': '__GFP_DMA32'}, 'GFP_HIGHUSER': {'value': 'GFP_USER | __GFP_HIGHMEM'}, 'GFP_HIGHUSER_MOVABLE': {'value': 'GFP_HIGHUSER | __GFP_MOVABLE'}, 'GFP_TRANSHUGE_LIGHT': {'value': 'GFP_HIGHUSER_MOVABLE | __GFP_COMP | __GFP_NOMEMALLOC | __GFP_NOWARN & ~__GFP_RECLAIM'}, 'GFP_TRANSHUGE': {'value': 'GFP_TRANSHUGE_LIGHT | __GFP_DIRECT_RECLAIM'}, '__GFP_DMA': {'value': 0x01}, '__GFP_HIGHMEM': {'value': 0x02}, '__GFP_DMA32': {'value': 0x04}, '__GFP_MOVABLE': {'value': 0x08}, '__GFP_RECLAIMABLE': {'value': 0x10}, '__GFP_HIGH': {'value': 0x20}, '__GFP_IO': {'value': 0x40}, '__GFP_FS': {'value': 0x80}, '__GFP_COLD': {'value': 0x100}, '__GFP_NOWARN': {'value': 0x200}, '__GFP_RETRY_MAYFAIL': {'value': 0x400}, '__GFP_NOFAIL': {'value': 0x800}, '__GFP_NORETRY': {'value': 0x1000}, '__GFP_MEMALLOC': {'value': 0x2000}, '__GFP_COMP': {'value': 0x4000}, '__GFP_ZERO': {'value': 0x8000}, '__GFP_NOMEMALLOC': {'value': 0x10000}, '__GFP_HARDWALL': {'value': 0x20000}, '__GFP_THISNODE': {'value': 0x40000}, '__GFP_ATOMIC': {'value': 0x80000}, '__GFP_ACCOUNT': {'value': 0x100000}, '__GFP_DIRECT_RECLAIM': {'value': 0x400000}, '__GFP_WRITE': {'value': 0x800000}, '__GFP_KSWAPD_RECLAIM': {'value': 0x1000000}, '__GFP_NOLOCKDEP': {'value': 0x2000000}, '__GFP_RECLAIM': {'value': '__GFP_DIRECT_RECLAIM|__GFP_KSWAPD_RECLAIM'}, } """ Definition of GFP flags The decimal value of a flag will be calculated by evaluating the entries from left to right. Grouping by parentheses is not supported. Source: include/linux/gpf.h @note : This list os probably a mixture of different kernel versions - be carefully @todo: Implement kernel specific versions because this flags are not constant (see https://github.com/torvalds/linux/commit/e67d4ca79aaf9d13a00d229b1b1c96b86828e8ba#diff-020720d0699e3ae1afb6fcd815ca8500) """ ps_table_items = ['pid', 'uid', 'tgid', 'total_vm_pages', 'rss_pages', 'nr_ptes_pages', 'swapents_pages', 'oom_score_adj'] """Elements of the process table""" class OOMEntity(object): """Hold whole OOM message block and provide access""" current_line = 0 """Zero based index of the current line in self.lines""" lines = [] """OOM text as list of lines""" state = "unknown" """State of the OOM after initial parsing""" text = "" """OOM as text""" def __init__(self, text): # use Unix LF only text = text.replace('\r\n', '\r') text = text.strip() oom_lines = text.split('\n') self.current_line = 0 self.lines = oom_lines self.text = text # don't do anything if the text is empty or does not contains the leading OOM message if not text: self.state = OOMEntityState.empty return elif 'invoked oom-killer:' not in text: self.state = OOMEntityState.invalid return oom_lines = self._remove_non_oom_lines(oom_lines) oom_lines = self._strip_needless_columns(oom_lines) oom_lines = self._rsyslog_unescape_lf(oom_lines) self.lines = oom_lines self.text = '\n'.join(oom_lines) if 'Killed process' in text: self.state = OOMEntityState.complete else: self.state = OOMEntityState.started def _number_of_columns_to_strip(self, first_line): """ Determinate number of columns left to the OOM message to strip. Sometime timestamps, hostnames and or syslog tags are left to the OOM message. This columns will be count to strip later. """ to_strip = 0 columns = first_line.split(" ") try: # strip all before " invoked oom-killer: gfp_mask=0x280da, order=0, oom_score_adj=0 to_strip = columns.index("invoked") # decrease to include to_strip -= 1 except ValueError: pass return to_strip def _remove_non_oom_lines(self, oom_lines): """Remove all lines before and after OOM message block""" cleaned_lines = [] in_oom_lines = False for line in oom_lines: # first line of the oom message block if "invoked oom-killer:" in line: in_oom_lines = True if in_oom_lines: cleaned_lines.append(line) # next line will not be part of the oom anymore if 'Killed process' in line: break return cleaned_lines def _rsyslog_unescape_lf(self, oom_lines): """ Rsyslog replaces line breaks with their ocal representation #012. This feature can be controlled inside the rsyslog configuration with the directives $EscapeControlCharactersOnReceive, $Escape8BitCharactersOnReceive and $ControlCharactersEscapePrefix. The replacement is only in second line (active_anon:....) of the Mem-Info block. """ lines = [] for line in oom_lines: if '#012' in line: line = line.replace('#012', '\r') expanded = line.split('\r') lines.extend(expanded) else: lines.append(line) return lines def _strip_needless_columns(self, oom_lines): """ Remove needless columns at the start of every line. This function removes all leading items w/o any relation to the OOM message like, date and time, hostname, syslog priority/facility. """ stripped_lines = [] cols_to_strip = self._number_of_columns_to_strip(oom_lines[0]) for line in oom_lines: # remove empty lines if not line.strip(): continue if cols_to_strip: # [-1] slicing needs Transcrypt operator overloading line = line.split(" ", cols_to_strip)[-1] # __:opov # OOMs logged to /var/log/messages / journalctl may contain # "kernel:" at the begin of the content if line.startswith('kernel:'): line = line[7:] stripped_lines.append(line) return stripped_lines def back(self): """Return the previous line""" if self.current_line - 1 < 0: raise StopIteration() self.current_line -= 1 return self.lines[self.current_line] def current(self): """Return the current line""" return self.lines[self.current_line] def next(self): """Return the next line""" if self.current_line + 1 < len(self.lines): self.current_line += 1 return self.lines[self.current_line] raise StopIteration() def find_text(self, pattern): """ Search the pattern and set the position to the first found line. Otherwise the position pointer won't be changed. :param pattern: Text to find :type pattern: str :return: True if the marker has found. """ for line in self.lines: if pattern in line: self.current_line = self.lines.index(line) return True return False def __iter__(self): return self def __next__(self): return self.next() class OOMAnalyser(object): """Analyse an OOM object and calculate additional values""" REC_INVOKED_OOMKILLER = re.compile( r'^(?P[\S ]+) invoked oom-killer: ' r'gfp_mask=(?P0x[a-z0-9]+)(\((?P[A-Z_|]+)\))?, ' r'(nodemask=(?P([\d,-]+|\(null\))), )?' r'order=(?P\d+), ' r'oom_score_adj=(?P\d+)', re.MULTILINE) REC_PID_KERNELVERSION = re.compile( r'^CPU: \d+ PID: (?P\d+) ' r'Comm: .* (Not tainted|Tainted:.*) ' r'(?P\d[\w.-]+) #\d', re.MULTILINE ) # split caused by a limited number of iterations during converting PY regex into JS regex REC_MEMINFO_1 = re.compile( # head line r'^Mem-Info:.*' # first line break r'(?:\n)' # first line (starting with a space) r'^active_anon:(?P\d+) inactive_anon:(?P\d+) ' r'isolated_anon:(?P\d+)' # next line break r'(?:\n)' # remaining lines (with leading space) r'^ active_file:(?P\d+) inactive_file:(?P\d+) ' r'isolated_file:(?P\d+)' # next line break r'(?:\n)' r'^ unevictable:(?P\d+) dirty:(?P\d+) writeback:(?P\d+) ' r'unstable:(?P\d+)' # # next line break # r'(?:\n)' # , re.MULTILINE ) REC_MEMINFO_2 = re.compile( r'^ slab_reclaimable:(?P\d+) slab_unreclaimable:(?P\d+)' r'(?:\n)' r'^ mapped:(?P\d+) shmem:(?P\d+) pagetables:(?P\d+) ' r'bounce:(?P\d+)' r'(?:\n)' r'^ free:(?P\d+) free_pcp:(?P\d+) free_cma:(?P\d+)', re.MULTILINE ) REC_MEM_NODEINFO = re.compile(r'(^Node \d+ (DMA|Normal|hugepages).*(:?\n))+', re.MULTILINE) REC_PAGECACHE = re.compile(r'^(?P\d+) total pagecache pages.*$', re.MULTILINE) REC_SWAP = re.compile( r'^(?P\d+) pages in swap cache' r'(?:\n)' r'^Swap cache stats: add \d+, delete \d+, find \d+/\d+' r'(?:\n)' r'^Free swap = (?P\d+)kB' r'(?:\n)' r'^Total swap = (?P\d+)kB', re.MULTILINE) REC_PAGEINFO = re.compile( r'^(?P\d+) pages RAM' r'(' r'(?:\n)' r'^(?P\d+) pages HighMem/MovableOnly' r')?' r'(?:\n)' r'^(?P\d+) pages reserved' r'(' r'(?:\n)' r'^(?P\d+) pages cma reserved' r')?' r'(' r'(?:\n)' r'^(?P\d+) pages in pagetable cache' r')?' r'(' r'(?:\n)' r'^(?P\d+) pages hwpoisoned' r')?', re.MULTILINE) REC_PROCESS_LINE = re.compile( r'^\[(?P[ \d]+)\]\s+(?P\d+)\s+(?P\d+)\s+(?P\d+)\s+(?P\d+)\s+' r'(?P\d+)\s+(?P\d+)\s+(?P-?\d+)\s+(?P.+)\s*') REC_OOM_KILL_PROCESS = re.compile( r'^Out of memory: Kill process (?P\d+) \((?P[\S ]+)\) ' r'score (?P\d+) or sacrifice child', re.MULTILINE ) REC_KILLED_PROCESS = re.compile( r'^Killed process \d+ \(.*\)' r'(, UID \d+,)?' r' total-vm:(?P\d+)kB, anon-rss:(?P\d+)kB, ' r'file-rss:(?P\d+)kB, shmem-rss:(?P\d+)kB.*', re.MULTILINE) lines = [] """All lines of an OOM without leading timestamps""" kernel_cfg = KernelConfig() """Kernel configuration""" results = {} """Extracted result""" oom_entity = None """Reference to the OOMEntity object""" def __init__(self, oom): self.results = {} self.oom_entity = oom def _extract_block_from_next_pos(self, marker): """ Extract a block starting with the marker and add all lines with a leading space character :rtype: str """ block = '' if not self.oom_entity.find_text(marker): return block line = self.oom_entity.current() block += "{}\n".format(line) for line in self.oom_entity: if not line.startswith(' '): self.oom_entity.back() break block += "{}\n".format(line) return block def _extract_from_oom_text(self): """Extract details from OOM message text""" self.results = {} for rec in [self.REC_INVOKED_OOMKILLER, self.REC_KILLED_PROCESS, self.REC_MEMINFO_1, self.REC_MEMINFO_2, self.REC_OOM_KILL_PROCESS, self.REC_PAGECACHE, self.REC_PAGEINFO, self.REC_PID_KERNELVERSION, self.REC_SWAP, ]: match = rec.search(self.oom_entity.text) if match: self.results.update(match.groupdict()) match = self.REC_MEM_NODEINFO.search(self.oom_entity.text) if match: self.results['mem_node_info'] = match.group() self.results['hardware_info'] = self._extract_block_from_next_pos('Hardware name:') # strip "Call Trace" line at beginning and remove leading spaces call_trace = '' block = self._extract_block_from_next_pos('Call Trace:') for line in block.split('\n'): if line.startswith('Call Trace'): continue call_trace += "{}\n".format(line.strip()) self.results['call_trace'] = call_trace # extract process table self.results['_ps'] = {} self.oom_entity.find_text('[ pid ]') for line in self.oom_entity: if not line.startswith('['): break if line.startswith('[ pid ]'): continue match = self.REC_PROCESS_LINE.match(line) if match: details = match.groupdict() details['notes'] = '' pid = details.pop('pid') self.results['_ps'][pid] = {} self.results['_ps'][pid].update(details) def _hex2flags(self, hexvalue, flag_definition): """\ Convert the hexadecimal value into flags specified by definition @return: list of flags and the decimal sum of all unknown flags """ remaining = int(hexvalue, 16) converted_flags = [] # __pragma__ ('jsiter') for flag in flag_definition: value = self._flag2decimal(flag, flag_definition) if remaining & value: # delete flag by "and" with a reverted mask remaining &= ~value converted_flags.append(flag) # __pragma__ ('nojsiter') return converted_flags, remaining def _flag2decimal(self, flag, flag_definition): """\ Convert a single flag into a decimal value """ if flag not in flag_definition: error('No definition for flag {} found'.format(flag)) return 0 value = flag_definition[flag]['value'] if isinstance(value, int): return value tokenlist = iter(re.split('([|&])', value)) operator = None negate_rvalue = False lvalue = 0 while True: try: token = next(tokenlist) except StopIteration: break token = token.strip() if token in ['|', '&']: operator = token continue if token.startswith('~'): token = token[1:] negate_rvalue = True if token.isdigit(): rvalue = int(token) elif token.startswith('0x') and token[2:].isdigit(): rvalue = int(token, 16) else: # it's not a decimal nor a hexadecimal value - reiterate assuming it's a flag string rvalue = self._flag2decimal(token, flag_definition) if negate_rvalue: rvalue = ~rvalue if operator == '|': lvalue |= rvalue elif operator == '&': lvalue &= rvalue operator = None negate_rvalue = False return lvalue def _convert_numeric_results_to_integer(self): """Convert all *_pages and *_kb to integer""" # __pragma__ ('jsiter') for item in self.results: if self.results[item] is None: self.results[item] = '' continue if item.endswith('_kb') or item.endswith('_pages') or item.endswith('_pid') or \ item in ['killed_proc_score', 'trigger_proc_order', 'trigger_proc_oomscore']: try: self.results[item] = int(self.results[item]) except: error('Converting item "{}={}" to integer failed'.format(item, self.results[item])) # __pragma__ ('nojsiter') def _convert_numeric_process_values_to_integer(self): """Convert numeric values in process table to integer values""" ps = self.results['_ps'] ps_index = [] # TODO Check if transcrypt issue: pragma jsiter for the whole block "for pid_str in ps: ..." # sets item in "for item in ['uid',..." to 0 instead of 'uid' # jsiter is necessary to iterate over ps for pid_str in ps.keys(): converted = {} process = ps[pid_str] for item in self.kernel_cfg.ps_table_items: if item == 'pid': continue try: converted[item] = int(process[item]) except: error('Converting process parameter "{}={}" to integer failed'.format(item, process[item])) converted['name'] = process['name'] converted['notes'] = process['notes'] pid_int = int(pid_str) del ps[pid_str] ps[pid_int] = converted ps_index.append(pid_int) ps_index.sort(key=int) self.results['_ps_index'] = ps_index def _calc_pstable_values(self): """Set additional notes to processes listed in the process table""" tpid = self.results['trigger_proc_pid'] kpid = self.results['killed_proc_pid'] # sometimes the trigger process isn't part of the process table if tpid in self.results['_ps']: self.results['_ps'][tpid]['notes'] = 'trigger process' # assume the killed process may also not part of the process table if kpid in self.results['_ps']: self.results['_ps'][kpid]['notes'] = 'killed process' def _calc_trigger_process_values(self): """Calculate all values related with the trigger process""" self.results['trigger_proc_requested_memory_pages'] = 2 ** self.results['trigger_proc_order'] self.results['trigger_proc_requested_memory_pages_kb'] = self.results['trigger_proc_requested_memory_pages'] * \ self.results['page_size_kb'] # process gfp_mask if self.results['trigger_proc_gfp_flags'] != '': # None has been is converted to '' flags = self.results['trigger_proc_gfp_flags'] del self.results['trigger_proc_gfp_flags'] else: flags, unknown = self._hex2flags(self.results['trigger_proc_gfp_mask'], self.kernel_cfg.GFP_FLAGS) if unknown: flags.append('0x{0:x}'.format(unknown)) flags = ' | '.join(flags) self.results['trigger_proc_gfp_mask'] = '{} ({})'.format(self.results['trigger_proc_gfp_mask'], flags) # already fully processed and no own element to display -> delete otherwise an error msg will be shown del self.results['trigger_proc_gfp_flags'] def _calc_killed_process_values(self): """Calculate all values related with the killed process""" self.results['killed_proc_total_rss_kb'] = self.results['killed_proc_anon_rss_kb'] + \ self.results['killed_proc_file_rss_kb'] + \ self.results['killed_proc_shmem_rss_kb'] self.results['killed_proc_rss_percent'] = int(100 * self.results['killed_proc_total_rss_kb'] / int(self.results['system_total_ram_kb'])) def _calc_swap_values(self): """Calculate all swap related values""" self.results['swap_cache_kb'] = self.results['swap_cache_pages'] * self.results['page_size_kb'] del self.results['swap_cache_pages'] # SwapUsed = SwapTotal - SwapFree - SwapCache self.results['swap_used_kb'] = self.results['swap_total_kb'] - self.results['swap_free_kb'] - \ self.results['swap_cache_kb'] self.results['system_swap_used_percent'] = int(100 * self.results['swap_total_kb'] / self.results['swap_used_kb']) def _calc_system_values(self): """Calculate system memory""" # educated guess self.results['page_size_kb'] = 4 # calculate remaining explanation values self.results['system_total_ram_kb'] = self.results['ram_pages'] * self.results['page_size_kb'] self.results['system_total_ramswap_kb'] = self.results['system_total_ram_kb'] + self.results['swap_total_kb'] total_rss_pages = 0 for pid in self.results['_ps'].keys(): total_rss_pages += self.results['_ps'][pid]['rss_pages'] self.results['system_total_ram_used_kb'] = total_rss_pages * self.results['page_size_kb'] self.results['system_total_used_percent'] = int(100 * self.results['system_total_ram_used_kb'] / self.results['system_total_ram_kb']) def _determinate_platform_and_distribution(self): """Determinate platform and distribution""" kernel_version = self.results.get('kernel_version', '') if 'x86_64' in kernel_version: self.results['platform'] = 'x86 64bit' else: self.results['platform'] = 'unknown' dist = 'unknown' if '.el7uek' in kernel_version: dist = 'Oracle Linux 7 (Unbreakable Enterprise Kernel)' elif '.el7' in kernel_version: dist = 'RHEL 7/CentOS 7' elif '.el6' in kernel_version: dist = 'RHEL 6/CentOS 6' elif '.el5' in kernel_version: dist = 'RHEL 5/CentOS 5' elif 'ARCH' in kernel_version: dist = 'Arch Linux' elif '_generic' in kernel_version: dist = 'Ubuntu' self.results['dist'] = dist def _calc_from_oom_details(self): """ Calculate values from already extracted details @see: self.results """ self._convert_numeric_results_to_integer() self._convert_numeric_process_values_to_integer() self._calc_pstable_values() self._determinate_platform_and_distribution() self._calc_system_values() self._calc_trigger_process_values() self._calc_killed_process_values() self._calc_swap_values() def analyse(self): """Extract and calculate values from the given OOM object""" self._extract_from_oom_text() self._calc_from_oom_details() return self.results class OOMDisplay(object): """Display the OOM analysis""" kernel_cfg = KernelConfig() """Kernel configuration""" oom_details = {} """Extracted result""" example = u'''\ sed invoked oom-killer: gfp_mask=0x201da, order=0, oom_score_adj=0 sed cpuset=/ mems_allowed=0-1 CPU: 4 PID: 29481 Comm: sed Not tainted 3.10.0-514.6.1.el7.x86_64 #1 Hardware name: HP ProLiant DL385 G7, BIOS A18 12/08/2012 ffff880182272f10 00000000021dcb0a ffff880418207938 ffffffff816861ac ffff8804182079c8 ffffffff81681157 ffffffff810eab9c ffff8804182fe910 ffff8804182fe928 0000000000000202 ffff880182272f10 ffff8804182079b8 Call Trace: [] dump_stack+0x19/0x1b [] dump_header+0x8e/0x225 [] ? ktime_get_ts64+0x4c/0xf0 [] ? delayacct_end+0x8f/0xb0 [] oom_kill_process+0x24e/0x3c0 [] ? oom_unkillable_task+0xcd/0x120 [] ? find_lock_task_mm+0x56/0xc0 [] ? has_capability_noaudit+0x1e/0x30 [] out_of_memory+0x4b6/0x4f0 [] __alloc_pages_slowpath+0x5d7/0x725 [] __alloc_pages_nodemask+0x405/0x420 [] alloc_pages_current+0xaa/0x170 [] __page_cache_alloc+0x97/0xb0 [] filemap_fault+0x170/0x410 [] ext4_filemap_fault+0x36/0x50 [ext4] [] __do_fault+0x4c/0xc0 [] do_read_fault.isra.42+0x43/0x130 [] handle_mm_fault+0x6b1/0xfe0 [] ? do_mmap_pgoff+0x305/0x3c0 [] __do_page_fault+0x154/0x450 [] do_page_fault+0x35/0x90 [] page_fault+0x28/0x30 Mem-Info: active_anon:7355653 inactive_anon:660960 isolated_anon:0 active_file:1263 inactive_file:1167 isolated_file:32 unevictable:0 dirty:4 writeback:0 unstable:0 slab_reclaimable:27412 slab_unreclaimable:13708 mapped:4818 shmem:87896 pagetables:25222 bounce:0 free:39513 free_pcp:2958 free_cma:0 Node 0 DMA free:15872kB min:40kB low:48kB high:60kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15992kB managed:15908kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? yes lowmem_reserve[]: 0 2780 15835 15835 Node 0 DMA32 free:59728kB min:7832kB low:9788kB high:11748kB active_anon:2154380kB inactive_anon:604748kB active_file:500kB inactive_file:112kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3094644kB managed:2848912kB mlocked:0kB dirty:0kB writeback:0kB mapped:4016kB shmem:5140kB slab_reclaimable:6448kB slab_unreclaimable:2796kB kernel_stack:1040kB pagetables:6876kB unstable:0kB bounce:0kB free_pcp:3788kB local_pcp:228kB free_cma:0kB writeback_tmp:0kB pages_scanned:28 all_unreclaimable? no lowmem_reserve[]: 0 0 13055 13055 Node 0 Normal free:36692kB min:36784kB low:45980kB high:55176kB active_anon:12301636kB inactive_anon:793132kB active_file:604kB inactive_file:176kB unevictable:0kB isolated(anon):0kB isolated(file):128kB present:13631488kB managed:13368348kB mlocked:0kB dirty:0kB writeback:0kB mapped:4108kB shmem:207940kB slab_reclaimable:47900kB slab_unreclaimable:28884kB kernel_stack:6624kB pagetables:43340kB unstable:0kB bounce:0kB free_pcp:4204kB local_pcp:640kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no lowmem_reserve[]: 0 0 0 0 Node 1 Normal free:49436kB min:45444kB low:56804kB high:68164kB active_anon:14967844kB inactive_anon:1244560kB active_file:1552kB inactive_file:1992kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:16777212kB managed:16514220kB mlocked:0kB dirty:16kB writeback:0kB mapped:10760kB shmem:138504kB slab_reclaimable:55300kB slab_unreclaimable:23152kB kernel_stack:6176kB pagetables:50672kB unstable:0kB bounce:0kB free_pcp:3360kB local_pcp:248kB free_cma:0kB writeback_tmp:0kB pages_scanned:125777 all_unreclaimable? yes lowmem_reserve[]: 0 0 0 0 Node 0 DMA: 0*4kB 0*8kB 0*16kB 0*32kB 2*64kB (U) 1*128kB (U) 1*256kB (U) 0*512kB 1*1024kB (U) 1*2048kB (M) 3*4096kB (M) = 15872kB Node 0 DMA32: 203*4kB (UEM) 231*8kB (UEM) 259*16kB (UEM) 231*32kB (UEM) 157*64kB (UEM) 90*128kB (UEM) 49*256kB (UEM) 20*512kB (UE) 3*1024kB (UEM) 1*2048kB (M) 0*4096kB = 63668kB Node 0 Normal: 1231*4kB (UEM) 391*8kB (UEM) 456*16kB (UEM) 342*32kB (UEM) 141*64kB (UEM) 23*128kB (UEM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 38260kB Node 1 Normal: 2245*4kB (UEM) 732*8kB (UEM) 594*16kB (UEM) 396*32kB (UEM) 160*64kB (UEM) 16*128kB (UEM) 2*256kB (UM) 0*512kB 1*1024kB (M) 0*2048kB 0*4096kB = 50836kB Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB Node 1 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB 100155 total pagecache pages 11342 pages in swap cache Swap cache stats: add 31260615, delete 31249273, find 295999950/297583545 Free swap = 0kB Total swap = 8388604kB 8379834 pages RAM 0 pages HighMem/MovableOnly 192987 pages reserved [ pid ] uid tgid total_vm rss nr_ptes swapents oom_score_adj name [ 390] 0 390 39012 6739 78 51 0 systemd-journal [ 433] 0 433 11104 2 22 360 -1000 systemd-udevd [ 530] 0 530 13854 28 27 83 -1000 auditd [ 559] 0 559 7692 65 19 87 0 systemd-logind [ 563] 0 563 4817 41 14 36 0 irqbalance [ 569] 87 569 7684 52 20 48 -900 dbus-daemon [ 587] 32 587 16240 17 34 116 0 rpcbind [ 647] 0 647 50303 11 36 113 0 gssproxy [ 796] 0 796 193856 2897 207 112 0 rsyslogd [ 818] 0 818 13177 0 27 146 0 vsftpd [ 840] 0 840 62892 9 36 103 0 ypbind [ 868] 0 868 21663 28 43 191 -1000 sshd [ 871] 29 871 11126 2 25 222 0 rpc.statd [ 907] 0 907 8044 4 21 53 0 atd [ 916] 0 916 27509 2 10 30 0 agetty [ 934] 0 934 27509 2 10 31 0 agetty [ 1255] 0 1255 45716 1 39 337 0 rscd [ 1268] 0 1268 45746 28 38 353 0 rscd [ 1269] 0 1269 45716 29 38 311 0 rscd [ 1285] 0 1285 23290 25 45 235 0 master [ 1287] 89 1287 23379 52 47 242 0 qmgr [ 1830] 0 1830 446643 959 68 1234 0 ovcd [ 2062] 0 2062 144894 511 37 309 0 ovbbccb [ 2121] 0 2121 33138 26 19 138 0 crond [ 2136] 38 2136 7846 40 19 88 0 ntpd [ 2451] 0 2451 177827 0 36 816 0 ovconfd [ 8145] 0 8145 300303 1616 58 692 0 hpsensor [ 8204] 0 8204 31508 119 31 328 0 opcmsgi [ 8405] 0 8405 201479 1289 49 244 0 opcmsga [ 8472] 0 8472 134080 236 46 514 0 opcmona [ 8596] 0 8596 31377 172 29 301 0 opcle [ 8658] 0 8658 81199 124 34 336 0 opcacta [ 8685] 0 8685 137169 23313 97 3256 0 oacore [ 6330] 12345 6330 7520 15 18 61 0 rotatelogs [ 6331] 12345 6331 28318 0 12 83 0 run.sh [ 6576] 12345 6576 8478546 5157063 15483 1527848 0 java [27171] 12345 27171 7522 10 18 58 0 rotatelogs [27172] 12345 27172 28320 3 11 94 0 run.sh [27502] 12345 27502 4029300 2716569 6505 226225 0 java [11729] 0 11729 64122 5003 79 2465 0 snmpd [12130] 0 12130 122202 565 29 175 0 hpasmlited [12166] 0 12166 11905 89 24 121 0 cmahealthd [12190] 0 12190 11871 89 24 119 0 cmastdeqd [12214] 0 12214 13707 84 31 211 0 cmahostd [12237] 0 12237 12493 38 28 352 0 cmathreshd [12276] 0 12276 12368 45 30 210 0 cmasm2d [12299] 0 12299 12485 43 26 282 0 cmaperfd [12324] 0 12324 31932 184 31 143 0 cmapeerd [12352] 0 12352 14280 48 32 169 0 cmaeventd [12379] 0 12379 14831 26 30 198 0 cmafcad [12407] 0 12407 11806 12 25 128 0 cmasasd [12436] 0 12436 14364 86 31 181 0 cmaidad [12463] 0 12463 11288 15 25 125 0 cmaided [12492] 0 12492 11805 14 26 127 0 cmascsid [12523] 0 12523 92228 129 63 433 0 cmanicd [14002] 0 14002 11803 12 25 128 0 cmasm2d [32615] 0 32615 36254 323 73 7 0 sshd [ 894] 12345 894 36254 328 70 5 0 sshd [ 895] 12345 895 3389 123 11 0 0 ksh [10620] 0 10620 36254 328 72 0 0 sshd [10634] 38714 10634 36290 329 70 8 0 sshd [10635] 38714 10635 14221 25 31 124 0 sftp-server [29021] 0 29021 36254 314 69 0 0 sshd [29025] 12345 29025 36254 316 67 0 0 sshd [29026] 12345 29026 29286 96 12 1 0 ksh [29051] 12345 29051 29494 330 12 74 0 svr05 [29979] 12345 29979 1666 42 9 0 0 less [29662] 89 29662 23316 258 43 0 0 pickup [26065] 89 26065 23317 256 45 0 0 trivial-rewrite [26066] 89 26066 23353 265 45 0 0 cleanup [26067] 89 26067 23368 271 45 0 0 smtp [26743] 0 26743 36254 314 68 0 0 sshd [26937] 12345 26937 36254 314 67 0 0 sshd [26938] 12345 26938 29286 96 11 0 0 ksh [27122] 12345 27122 29494 459 12 0 0 svr05 [28657] 0 28657 36254 314 74 0 0 sshd [28702] 12345 28702 36254 314 72 0 0 sshd [28703] 12345 28703 29286 97 11 0 0 ksh [28993] 0 28993 36254 314 72 0 0 sshd [28996] 12345 28996 29526 531 12 0 0 svr05 [29006] 12345 29006 36254 314 69 0 0 sshd [29007] 12345 29007 29286 96 11 0 0 ksh [29110] 12345 29110 29558 745 12 0 0 svr05 [29481] 12345 29481 29214 58 14 0 0 sed [29752] 12345 29752 7522 296 19 0 0 rotatelogs Out of memory: Kill process 6576 (java) score 651 or sacrifice child Killed process 6576 (java) total-vm:33914892kB, anon-rss:20629004kB, file-rss:0kB, shmem-rss:0kB ''' sorted_column = None """Processes will sort by values in this column""" sort_order = None """Sort order for process values""" svg_namespace = 'http://www.w3.org/2000/svg' # from Sasha Trubetskoy - https://sashat.me/2017/01/11/list-of-20-simple-distinct-colors/ svg_colours = [ '#e6194b', # Red '#3cb44b', # Green '#ffe119', # Yellow '#0082c8', # Blue '#f58231', # Orange '#911eb4', # Purple '#46f0f0', # Cyan '#f032e6', # Magenta '#d2f53c', # Lime '#fabebe', # Pink '#008080', # Teal '#e6beff', # Lavender '#aa6e28', # Brown '#fffac8', # Beige '#800000', # Maroon '#aaffc3', # Mint '#808000', # Olive '#ffd8b1', # Coral '#000080', # Navy '#808080', # Grey ] svg_array_updown = """ """ """SVG graphics with two black triangles UP and DOWN for sorting""" svg_array_up = """ """ """SVG graphics with one black triangle UP for sorting""" svg_array_down = """ """ """SVG graphics with one black triangle DOWN for sorting""" def __init__(self): self.set_HTML_defaults() self.update_toc() element = document.getElementById('version') element.textContent = "v{}".format(VERSION) def _set_item(self, item): """ Paste the content into HTML elements with the ID / Class that matches the item name. The content won't be formatted. Only suffixes for pages and kbytes are added in the singular or plural. """ elements = document.getElementsByClassName(item) for element in elements: content = self.oom_details.get(item, '') if isinstance(content, str): content = content.strip() if content == '': row = element.parentNode row.classList.add('js-text--display-none') if item.endswith('_pages') and isinstance(content, int): if content == 1: content = "{} page".format(content) else: content = "{} pages".format(content) if item.endswith('_kb') and isinstance(content, int): if content == 1: content = "{} kByte".format(content) else: content = "{} kBytes".format(content) if item.endswith('_percent') and isinstance(content, int): content = "{}%".format(content) element.textContent = content if DEBUG: show_element('notify_box') def update_toc(self): """ Update the TOC to show current headlines only There are two conditions to show a h2 headline in TOC: * the headline is visible * the id attribute is set """ new_toc = '' toc_content = document.querySelectorAll('nav > ul')[0] for element in document.querySelectorAll('h2'): if not (is_visible(element) and element.id): continue new_toc += '
  • {}
    • '.format(element.id, element.textContent) toc_content.innerHTML = new_toc def update_process_table(self): """ Re-create the process table with additional information """ new_table = '' table_content = document.getElementById('process_table') for pid in self.oom_details['_ps_index']: if pid == self.oom_details['trigger_proc_pid']: css_class = 'class="js-pstable__triggerproc--bgcolor"' elif pid == self.oom_details['killed_proc_pid']: css_class = 'class="js-pstable__killedproc--bgcolor"' else: css_class = '' process = self.oom_details['_ps'][pid] line = """ {} {} {} {} {} {} {} {} {} {} """.format(css_class, pid, process['uid'], process['tgid'], process['total_vm_pages'], process['rss_pages'], process['nr_ptes_pages'], process['swapents_pages'], process['oom_score_adj'], process['name'], process['notes']) new_table += line table_content.innerHTML = new_table self.set_sort_triangle() def set_sort_triangle(self): """Set the sorting symbols for all columns in the process table""" # TODO Check operator overloading # Operator overloading (Pragma opov) does not work in this context. # self.kernel_cfg.ps_table_items + ['notes'] will compile to a string # "pid,uid,tgid,total_vm_pages,rss_pages,nr_ptes_pages,swapents_pages,oom_score_adjNotes" and not to an # array ps_table_and_notes = self.kernel_cfg.ps_table_items[:] ps_table_and_notes.append('notes') for column_name in ps_table_and_notes: element_id = "pstable_sort_{}".format(column_name) element = document.getElementById(element_id) if not element: internal_error('Missing id "{}" in process table.'.format(element_id)) continue if column_name == self.sorted_column: if self.sort_order == 'descending': element.innerHTML = self.svg_array_down else: element.innerHTML = self.svg_array_up else: element.innerHTML = self.svg_array_updown def set_HTML_defaults(self): """Reset the HTML document but don't clean elements""" # hide all elements marked to be hidden by default for element in document.querySelectorAll('.js-text--default-hide'): element.classList.add('js-text--display-none') # show all elements marked to be shown by default for element in document.querySelectorAll('.js-text--default-show'): element.classList.remove('js-text--display-none') # show hidden rows for element in document.querySelectorAll('table .js-text--display-none'): element.classList.remove('js-text--display-none') # clear notification box element = document.getElementById('notify_box') while element.firstChild: element.removeChild(element.firstChild) # clear process table element = document.getElementById('process_table') while element.firstChild: element.removeChild(element.firstChild) # reset sort triangles self.sorted_column = None self.sort_order = None self.set_sort_triangle() # remove svg charts for element_id in ('svg_swap', 'svg_ram'): element = document.getElementById(element_id) while element.firstChild: element.removeChild(element.firstChild) def svg_create_element(self, height, width, css_class): """Return an empty SVG element""" svg = document.createElementNS(self.svg_namespace, 'svg') svg.setAttribute('version', '1.1') svg.setAttribute('height', height) svg.setAttribute('width', width) svg.setAttribute('viewBox', '0 0 {} {}'.format(width, height)) svg.setAttribute('class', css_class) return svg def svg_create_rect(self, x=0, y=0, width=0, height=0, colour=None): rect = document.createElementNS(self.svg_namespace, 'rect') if x: rect.setAttribute('x', x) if y: rect.setAttribute('y', y) if width: rect.setAttribute('width', width) if height: rect.setAttribute('height', height) if colour: rect.setAttribute('fill', colour) return rect def svg_generate_bar_chart(self, css_class, *elements): """Generate a SVG bar chart""" bar_height = 100 label_height = 80 length_factor = 4 overall_height = bar_height + label_height overall_width = 100 * length_factor css_class = 'js-mem-usage__svg' svg = self.svg_create_element(overall_height, overall_width, css_class) sum_all_elements = sum([length for unused, length in elements]) current_pos = 0 bar_group = document.createElementNS(self.svg_namespace, 'g') bar_group.setAttribute('id', 'bar_group') bar_group.setAttribute('stroke', 'black') bar_group.setAttribute('stroke-width', 2) nr_processed_elements = 0 for title, length in elements: # length is None/undefined is the regular expression doesn't find any values if not length: continue rect_len = int(100 * length / sum_all_elements) * length_factor if not rect_len: continue colour = self.svg_colours[nr_processed_elements % len(self.svg_colours)] rect = self.svg_create_rect(current_pos, 0, rect_len, bar_height, colour) bar_group.appendChild(rect) label_group = document.createElementNS(self.svg_namespace, 'g') label_group.setAttribute('id', title) colour_rect = self.svg_create_rect(0, 0, 20, 20, colour) colour_rect.setAttribute('stroke', 'black') colour_rect.setAttribute('stroke-width', 2) text = document.createElementNS(self.svg_namespace, 'text') text.setAttribute('x', '30') text.setAttribute('y', '18') text.textContent = title label_group.appendChild(colour_rect) label_group.appendChild(text) # TODO replace hardcoded values x = 5 + 125 * (nr_processed_elements // 2) y = bar_height + 10 + (nr_processed_elements % 2) * 40 label_group.setAttribute('transform', 'translate({}, {})'.format(x, y)) bar_group.appendChild(label_group) current_pos += rect_len nr_processed_elements += 1 svg.appendChild(bar_group) return svg def copy_example_to_form(self): document.getElementById('textarea_oom').value = self.example def reset_form(self): document.getElementById('textarea_oom').value = "" self.set_HTML_defaults() self.update_toc() def toggle_oom(self, show=False): """Toggle the visibility of the full OOM message""" oom_element = document.getElementById('oom') row_with_oom = oom_element.parentNode.parentNode toggle_msg = document.getElementById('oom_toogle_msg') if show or row_with_oom.classList.contains('js-text--display-none'): row_with_oom.classList.remove('js-text--display-none') toggle_msg.text = "(click to hide)" else: row_with_oom.classList.add('js-text--display-none') toggle_msg.text = "(click to show)" def analyse_and_show(self): """Analyse the OOM text inserted into the form and show the results""" self.oom = OOMEntity(self.load_from_form()) if not self.is_valid(self.oom): self.oom = None return # set defaults and clear notifications self.oom_details.clear() self.set_HTML_defaults() # analyse analyser = OOMAnalyser(self.oom) self.oom_details = analyser.analyse() # Update kernel configuration self.kernel_cfg = analyser.kernel_cfg # display results self.show() self.update_toc() def load_from_form(self): element = document.getElementById('textarea_oom') oom_text = element.value return oom_text def is_valid(self, oom): """ Return True for a complete OOM otherwise False and a warning msg for a incomplete or an error msg if the start sequence was not found. """ if oom.state == OOMEntityState.complete: return True elif oom.state == OOMEntityState.started: warning('The inserted OOM is incomplete!') warning('The initial pattern was found but not the final. The result may be incomplete!') elif oom.state == OOMEntityState.invalid: error('The inserted text is not a valid OOM block!') error('The initial pattern was not found!') elif oom.state == OOMEntityState.empty: error('The inserted text is empty! Please insert an OOM message block.') else: error('Invalid state "{}" after the OOM has formally checked!'.format(self.oom.state)) return False def show(self): """ Show all extracted details as well as additionally generated information """ if DEBUG: print(self.oom_details) hide_element('input') show_element('analysis') for item in self.oom_details.keys(): # ignore internal items if item.startswith('_'): continue self._set_item(item) # generate process table self.update_process_table() # generate swap usage diagram svg_swap = self.svg_generate_bar_chart( ('Swap Used', self.oom_details['swap_used_kb']), ('Swap Free', self.oom_details['swap_free_kb']), ('Swap Cached', self.oom_details['swap_cache_kb']), ) elem_svg_swap = document.getElementById('svg_swap') elem_svg_swap.appendChild(svg_swap) # generate RAM usage diagram svg_ram = self.svg_generate_bar_chart( ('Active mem', self.oom_details['active_anon_pages']), ('Inactive mem', self.oom_details['inactive_anon_pages']), ('Isolated mem', self.oom_details['isolated_anon_pages']), ('Active PC', self.oom_details['active_file_pages']), ('Inactive PC', self.oom_details['inactive_file_pages']), ('Isolated PC', self.oom_details['isolated_file_pages']), ('Unevictable', self.oom_details['unevictable_pages']), ('Dirty', self.oom_details['dirty_pages']), ('Writeback', self.oom_details['writeback_pages']), ('Unstable', self.oom_details['unstable_pages']), ('Slab reclaimable', self.oom_details['slab_reclaimable_pages']), ('Slab unreclaimable', self.oom_details['slab_unreclaimable_pages']), ('Mapped', self.oom_details['mapped_pages']), ('Shared', self.oom_details['shmem_pages']), ('Pagetable', self.oom_details['pagetables_pages']), ('Bounce', self.oom_details['bounce_pages']), ('Free', self.oom_details['free_pages']), ('Free PCP', self.oom_details['free_pcp_pages']), ('Free CMA', self.oom_details['free_cma_pages']), ) elem_svg_ram = document.getElementById('svg_ram') elem_svg_ram.appendChild(svg_ram) element = document.getElementById('oom') element.textContent = self.oom.text self.toggle_oom(show=False) def sort_pstable(self, column_name): """Sort process table by the values in the given column""" # TODO Check operator overloading # Operator overloading (Pragma opov) does not work in this context. # self.kernel_cfg.ps_table_items + ['notes'] will compile to a string # "pid,uid,tgid,total_vm_pages,rss_pages,nr_ptes_pages,swapents_pages,oom_score_adjNotes" and not to an # array ps_table_and_notes = self.kernel_cfg.ps_table_items[:] ps_table_and_notes.append('notes') if column_name not in ps_table_and_notes: internal_error('Can not sort process table with an unknown column name "{}"'.format(column_name)) return # reset sort order if the column has changes if column_name != self.sorted_column: self.sort_order = None self.sorted_column = column_name if not self.sort_order or self.sort_order == 'descending': self.sort_order = 'ascending' self.sort_psindex_by_column(column_name) else: self.sort_order = 'descending' self.sort_psindex_by_column(column_name, True) self.sorted_column = column_name self.update_process_table() def sort_psindex_by_column(self, column_name, reverse=False): """ Sort the pid list '_ps_index' based on the values in the process dict '_ps'. Is uses bubble sort with all disadvantages but just a few lines of code """ ps = self.oom_details['_ps'] ps_index = self.oom_details['_ps_index'] def getvalue(column_name, i): if column_name == 'pid': value = ps_index[i] else: value = ps[ps_index[i]][column_name] # JS sorts alphanumeric by default, convert values explicit to integers to sort numerically if column_name not in ['name', 'notes'] and value is not js_undefined: value = int(value) return value # We set swapped to True so the loop looks runs at least once swapped = True while swapped: swapped = False for i in range(len(ps_index) - 1): v1 = getvalue(column_name, i) v2 = getvalue(column_name, i+1) if (not reverse and v1 > v2) or (reverse and v1 < v2): # Swap the elements ps_index[i], ps_index[i+1] = ps_index[i+1], ps_index[i] # Set the flag to True so we'll loop again swapped = True OOMDisplayInstance = OOMDisplay()