Checkmk git commits April 2019

checkmk-commits@lists.checkmk.com

15 participants
599 discussions

by Sergey Kipnis

Module: check_mk Branch: master Commit: 6b2a35f2853fa9a030725c025198c70f8a5d3059 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=6b2a35f2853fa9… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Thu Mar 28 19:55:07 2019 +0100 [CMK-1791] - memory leak in ps(WMI) Change-Id: Ib3664e7114c1ce688849f4c884e84193a2454a4c --- agents/wnx/src/engine/providers/ps.cpp | 1 + 1 file changed, 1 insertion(+) diff --git a/agents/wnx/src/engine/providers/ps.cpp b/agents/wnx/src/engine/providers/ps.cpp index 53554b3..3ad5ba5 100644 --- a/agents/wnx/src/engine/providers/ps.cpp +++ b/agents/wnx/src/engine/providers/ps.cpp @@ -266,6 +266,7 @@ std::string ProducePsWmi(bool FullPath) { XLOG::l("Skipping scanning, enumerator can't be opened"); return {}; } + ON_OUT_OF_SCOPE(processes->Release()); std::string out; while (1) {

5 years, 3 months

[CMK-1791] - small improvements

by Sergey Kipnis

Module: check_mk Branch: master Commit: a99c586c056aed3424a925194cbe3eb2a6e8597c URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=a99c586c056aed… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Thu Mar 28 19:55:49 2019 +0100 [CMK-1791] - small improvements - empty section in user config is allowed - minimum timeout set to 5 seconds to avoid random dropping WMI output Change-Id: Ia6c547583824d0ff773581bfad8ab64c1dada4bd --- agents/wnx/src/common/cfg_info.h | 2 +- agents/wnx/src/engine/cfg.cpp | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/agents/wnx/src/common/cfg_info.h b/agents/wnx/src/common/cfg_info.h index 1533f1e..cb342a7 100644 --- a/agents/wnx/src/common/cfg_info.h +++ b/agents/wnx/src/common/cfg_info.h @@ -62,7 +62,7 @@ constexpr int kDefaultWinPerfTimeout = 11; // seconds constexpr int kDefaultAgentMaxWait = 15; // max time agent waits for a sections // #TODO CONFIRM VALUE: -constexpr int kDefaultAgentMinWait = 1; // max time agent waits for a sections +constexpr int kDefaultAgentMinWait = 5; // max time agent waits for a sections const char* const kDefaultLogFileName = "check_mk.log"; const char* const kDefaultAppFileName = "check_mk_agent.exe"; diff --git a/agents/wnx/src/engine/cfg.cpp b/agents/wnx/src/engine/cfg.cpp index ed3415b..ca03eca 100644 --- a/agents/wnx/src/engine/cfg.cpp +++ b/agents/wnx/src/engine/cfg.cpp @@ -780,7 +780,8 @@ bool ConfigInfo::smartMerge(YAML::Node Target, const YAML::Node Src, smartMerge(grp, s, merge_seq); } } else { - XLOG::l.bp(XLOG_FLINE + " expected map from source {}", name); + if (s.IsNull()) continue; // empty skipped + XLOG::l(XLOG_FLINE + " expected map from source {}", name); } continue; } else if (IsYamlSeq(grp)) {

5 years, 3 months

[CMK-1890] - deploy script checks whether automatic update works

by Sergey Kipnis

Module: check_mk Branch: master Commit: 172b64dd3212c54b8901cfa0c12b4d762f412bc4 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=172b64dd3212c5… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Thu Mar 28 18:12:31 2019 +0100 [CMK-1890] - deploy script checks whether automatic update works Change-Id: I427cf15aa86438e6cdbc8fbf05b026c430e01560 --- agents/wnx/build_release.cmd | 16 ++++++++++++++++ agents/wnx/call_deploy.cmd | 24 ++++++++++++++++++------ 2 files changed, 34 insertions(+), 6 deletions(-) diff --git a/agents/wnx/build_release.cmd b/agents/wnx/build_release.cmd index 76aebc2..9dcda5c 100644 --- a/agents/wnx/build_release.cmd +++ b/agents/wnx/build_release.cmd @@ -59,8 +59,22 @@ if not %errorlevel% == 0 powershell Write-Host "Failed %exec%-32" -Foreground Re %msbuild% wamain.sln /t:%exec% /p:Configuration=Release,Platform=x64 if not %errorlevel% == 0 powershell Write-Host "Failed %exec%-64" -Foreground Red && exit 7 +@rem auto install msi +git update-index --assume-unchanged install/resources/check_mk.dat > nul +@copy install\resources\check_mk.dat save.tmp > nul +echo update > install\resources\check_mk.dat +%msbuild% wamain.sln /t:install /p:Configuration=Release,Platform=x64 +set el=%errorlevel% +@type save.tmp > install\resources\check_mk.dat +@del save.tmp > nul +git update-index --no-assume-unchanged install/resources/check_mk.dat > nul +if not %el% == 0 powershell Write-Host "Failed Install build" -Foreground Red && exit 88 +copy %REMOTE_MACHINE%\check_mk_service.msi %REMOTE_MACHINE%\check_mk_agent_update.msi + + %msbuild% wamain.sln /t:install /p:Configuration=Release,Platform=x64 if not %errorlevel% == 0 powershell Write-Host "Failed Install build" -Foreground Red && exit 8 + goto end @rem ignored: powershell Write-Host "starting unit tests" -Foreground Cyan @@ -83,5 +97,7 @@ copy %REMOTE_MACHINE%\check_mk_service.msi %REMOTE_MACHINE%\check_mk_agent.msi copy %REMOTE_MACHINE%\check_mk_service32.exe %REMOTE_MACHINE%\check_mk_agent.exe copy %REMOTE_MACHINE%\check_mk_service64.exe %REMOTE_MACHINE%\check_mk_agent-64.exe +rem touching update msi +copy %REMOTE_MACHINE%\check_mk_agent_update.msi /B+ ,,/Y > nul diff --git a/agents/wnx/call_deploy.cmd b/agents/wnx/call_deploy.cmd index 1558f09..8e4fa9a 100644 --- a/agents/wnx/call_deploy.cmd +++ b/agents/wnx/call_deploy.cmd @@ -1,7 +1,7 @@ @echo off powershell Write-Host Starting deployment -Foreground Green if "%1" == "SIMULATE_OK" powershell Write-Host "Deploy: SUCCESS" -Foreground Green && exit 0 -if "%1" == "SIMULATE_FAIL" powershell Write-Host "Deploy: FAIL" -Foreground Red && del %REMOTE_MACHINE%\check_mk_service.msi && exit 99 +if "%1" == "SIMULATE_FAIL" powershell Write-Host "Deploy: FAIL" -Foreground Red && del %REMOTE_MACHINE%\check_mk_agent.msi && exit 99 set root=%cd%\..\..\artefacts set REMOTE_MACHINE=%root% @set svc=CheckMkService @@ -9,25 +9,35 @@ set p="%ProgramFiles(X86)%\check_mk_service\upgrade.protocol" pushd cd %root% powershell Write-Host "%svc% is cleaning up" -Foreground DarkGray -msiexec /uninstall check_mk_service.msi /quiet /LV* log.tmp +msiexec /uninstall check_mk_agent.msi /quiet /LV* log.tmp del %p% 2> null powershell Write-Host "%svc% is installing" -Foreground Cyan -msiexec /I check_mk_service.msi /quiet /LV* log +msiexec /I check_mk_agent.msi /quiet /LV* log powershell Write-Host "%svc% is installed" -Foreground Green powershell Start-Sleep 2 @set status=1 powershell Write-Host "checking status" -Foreground Cyan sc query "%svc%" | find /i "RUNNING" > null -if %ERRORLEVEL% == 0 goto work +if %ERRORLEVEL% == 0 goto update powershell Write-Host "%svc% is NOT running" -Foreground Red set status=3 goto end +:update +@rem Update check +mkdir %ALLUSERSPROFILE%\CheckMk\Agent\Update 2> nul +copy check_mk_agent_update.msi %ALLUSERSPROFILE%\CheckMk\Agent\update\check_mk_agent.msi > nul +powershell Start-Sleep 20 +echo update > control.tmp +fc "%ProgramFiles(X86)%\check_mk_service\check_mk.dat" control.tmp > nul +set upd=%errorlevel% +del control.tmp > nul :work +@rem Uninstall check powershell Write-Host "%svc% is running" -Foreground Green if not exist %p% powershell Write-Host "Upgrade Protocol file not found" -Foreground Red && set status=5 && goto end powershell Write-Host "protocol file is found" -Foreground Green powershell Write-Host "%svc% is uninstalling" -Foreground Cyan -msiexec /uninstall check_mk_service.msi /quiet /LV* log.tmp +msiexec /uninstall check_mk_agent.msi /quiet /LV* log.tmp sc query "%svc%1" > null if %ERRORLEVEL% == 0 powershell Write-Host "Cannot Uninstall Service" -Foreground Red && set status=5 && goto end powershell Write-Host "%svc% is uninstalled" -Foreground Green @@ -35,4 +45,6 @@ set status=0 :end popd del %p% -exit /B %status% +if "%upd%" == "0" exit /B %status% +powershell Write-Host "%svc% is failed in update" -Foreground Red +exit /B %upd%

5 years, 3 months

[CMK-1791] - trace output

by Sergey Kipnis

Module: check_mk Branch: master Commit: cfac18369cf1787947b1a64ab7d86f82dd6dfb5a URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=cfac18369cf178… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Thu Mar 28 16:33:09 2019 +0100 [CMK-1791] - trace output Change-Id: I9d074f480c2dee863813962eb19b91ed274b141d --- agents/wnx/src/engine/async_answer.cpp | 9 +++------ agents/wnx/src/engine/carrier.cpp | 1 + agents/wnx/src/engine/service_processor.h | 6 ++++-- 3 files changed, 8 insertions(+), 8 deletions(-) diff --git a/agents/wnx/src/engine/async_answer.cpp b/agents/wnx/src/engine/async_answer.cpp index b29565c..5cc02a5 100644 --- a/agents/wnx/src/engine/async_answer.cpp +++ b/agents/wnx/src/engine/async_answer.cpp @@ -102,14 +102,13 @@ bool AsyncAnswer::addSegment( ) { std::lock_guard lk(lock_); if (Id != tp_id_) { - xlog::d("Invalid attempt to add data"); + XLOG::d("Invalid attempt to add data '{}'", SectionName); return false; } for (const auto& s : segments_) { if (s.name_ == SectionName) { - xlog::l("Section %s tries to store data twice. F-f", - SectionName.c_str()); + XLOG::l("Section '{}' tries to store data twice. F-f", SectionName); return false; // duplicated section run } } @@ -122,7 +121,7 @@ bool AsyncAnswer::addSegment( } } catch (const std::exception& e) { // not possible, but we have to check - xlog::l(XLOG_FLINE + "-exception %s", e.what()); + XLOG::l(XLOG_FLINE + "-exception '{}'", e.what()); } received_segments_++; @@ -136,8 +135,6 @@ bool AsyncAnswer::addSegment( return true; } - - // resets data, internal use only void AsyncAnswer::dropDataNoLock() { tp_id_ = GenerateAnswerId(); diff --git a/agents/wnx/src/engine/carrier.cpp b/agents/wnx/src/engine/carrier.cpp index 83e0237..9678aa0 100644 --- a/agents/wnx/src/engine/carrier.cpp +++ b/agents/wnx/src/engine/carrier.cpp @@ -80,6 +80,7 @@ bool CoreCarrier::establishCommunication(const std::string& InternalPort) { bool CoreCarrier::sendData(const std::string& PeerName, uint64_t Marker, const void* Data, size_t Length) { std::lock_guard lk(lock_); + XLOG::d.t("Sending data '{}' id is [{}] length [{}]", PeerName, Marker, Length); return sendDataDispatcher(DataType::kSegment, PeerName, Marker, Data, Length); } diff --git a/agents/wnx/src/engine/service_processor.h b/agents/wnx/src/engine/service_processor.h index 645a7e1..ae9d925 100644 --- a/agents/wnx/src/engine/service_processor.h +++ b/agents/wnx/src/engine/service_processor.h @@ -255,6 +255,8 @@ private: cma::OnStartApp(); auto tp = openAnswer(Ip); if (tp) { + XLOG::d.i("Id is [{}] ", + tp.value().time_since_epoch().count()); auto started = startProviders(tp.value(), Ip); return getAnswer(started); @@ -487,10 +489,10 @@ private: [this](const AnswerId Tp, ServiceProcessor* Proc) { auto block = gatherData(); if (block) { - XLOG::d("Provider {} added answer", name_); + XLOG::d("Provider '{}' added answer", name_); return Proc->addSectionToAnswer(name_, Tp, *block); } else { - XLOG::l("Provider {} FAILED answer", name_); + XLOG::l("Provider '{}' FAILED answer", name_); Proc->addSectionToAnswer(name_, Tp); return false; }

5 years, 3 months

[CMK-1791] - removing plugin player from the main thread. We do not need it any more. Stub left for the future

by Sergey Kipnis

Module: check_mk Branch: master Commit: 8113ab828ee227612718e137f2128c0e9fa3b3d9 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=8113ab828ee227… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Thu Mar 28 16:32:42 2019 +0100 [CMK-1791] - removing plugin player from the main thread. We do not need it any more. Stub left for the future Change-Id: I8445eaf79c2d9fcd42d58bd9b71811541456c167 --- agents/wnx/src/engine/service_processor.cpp | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/agents/wnx/src/engine/service_processor.cpp b/agents/wnx/src/engine/service_processor.cpp index 4b10a58..acaffba 100644 --- a/agents/wnx/src/engine/service_processor.cpp +++ b/agents/wnx/src/engine/service_processor.cpp @@ -229,10 +229,14 @@ int ServiceProcessor::startProviders(AnswerId Tp, std::string Ip) { } // Plugins Processing - if (0 && groups::plugins.enabledInConfig()) { +#if 0 + // we do not use anymore separate plugin process(player) + // code is here as future refernce to use again separate process + if (groups::plugins.enabledInConfig()) { cma::cfg::SetupPluginEnvironment(); kickPlugins(Tp, Ip); } +#endif if (max_timeout_ <= 0) { max_timeout_ = cma::cfg::kDefaultAgentMinWait;

5 years, 3 months

wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable

by Simon Betz

Module: check_mk Branch: master Commit: c28ced6fc07f156b7c6ec753a0b0e1d98e65d1c4 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=c28ced6fc07f15… Author: Simon Betz <si(a)mathias-kettner.de> Date: Mon Apr 1 13:49:41 2019 +0200 wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable Change-Id: Icb8ec4fb2cc22f654f9971821154995928759d55 --- checks/wmi.include | 111 +++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 83 insertions(+), 28 deletions(-) diff --git a/checks/wmi.include b/checks/wmi.include index bbf3420..2fea9ed 100644 --- a/checks/wmi.include +++ b/checks/wmi.include @@ -45,13 +45,6 @@ # '----------------------------------------------------------------------' -class NullTable(object): - """Represents an empty table caused by a WMI timeout.""" - - def __getattr__(self, name): - raise MKCounterWrapped('WMI query timed out') - - class WMITable(object): """ Represents a 2-dimensional table of performance metrics. @@ -72,10 +65,9 @@ class WMITable(object): self.__headers = {} self.__timestamp = timestamp self.__frequency = frequency - index = 0 prev_header = None - for header in headers: + for index, header in enumerate(headers): if not header.strip() and prev_header: # MS apparently doesn't bother to provide a name # for base columns with performance counters @@ -83,7 +75,7 @@ class WMITable(object): header = self._normalize_key(header) self.__headers[header] = index prev_header = header - index += 1 + self.__key_index = None if key_field is not None: try: @@ -94,6 +86,7 @@ class WMITable(object): self.__row_lookup = {} self.__rows = [] + self.timed_out = False if rows: for row in rows: self.add_row(row) @@ -129,9 +122,37 @@ class WMITable(object): if key in WMITable.TOTAL_NAMES: key = row[self.__key_index] = None self.__row_lookup[key] = len(self.__rows) + self.__rows.append(row) + if not self.timed_out: + # Check if there's a timeout in the last added line + # ie. row (index) == -1, column 'WMIStatus' + try: + wmi_status = self._get_row_col_value(-1, 'WMIStatus') + except IndexError: + #TODO Why does the agent send data with different length? + # Eg. skype + # tablename = [LS:WEB - EventChannel] + # header = [ + # u'instance', u'EventChannel - Pending Get', u' Timed Out Request Count', + # u'EventChannel - Pending Get', u' Active Request Count', u'EventChannel - Push Events', + # u' Channel Clients Active', u'EventChannel - Push Events', u' Channel Clients Disposed', + # u'EventChannel - Push Events', u' Notification Requests Sent', + # u'EventChannel - Push Events', u' Heartbeat Requests Sent', u'EventChannel - Push Events', + # u' Requests Succeeded', u'EventChannel - Push Events', u' Requests Failed' + # ] + # row = [u'"_Total"', u'259', u'1', u'0', u'0', u'0', u'0', u'0', u'0'] + # Then we try to check last value of row + wmi_status = self._get_row_col_value(-1, -1) + if wmi_status.lower() == "timeout": + self.timed_out = True def get(self, row, column): + if self.timed_out: + raise MKCounterWrapped('WMI query timed out') + return self._get_row_col_value(row, column) + + def _get_row_col_value(self, row, column): if isinstance(row, int): row_index = row else: @@ -145,7 +166,6 @@ class WMITable(object): except KeyError, e: raise MKGeneralException(e.message + " missing, valid keys: " + ", ".join(self.__headers.keys())) - return self.__rows[row_index][col_index] def row_labels(self): @@ -198,23 +218,15 @@ def parse_wmi_table(info, key="Name"): # single-table input tablename = "" - if is_wmi_timeout(line): - parsed[tablename] = NullTable() - line = info_iter.next() - continue - - if tablename not in parsed: - parsed[tablename] = WMITable(tablename, line, key, timestamp, frequency) - current = parsed[tablename] + missing_wmi_status, current_table =\ + _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency) # read table content line = info_iter.next() while not line[0].startswith("["): - if is_wmi_timeout(line): - parsed[tablename] = NullTable() - line = info_iter.next() - continue - current.add_row(line) + if missing_wmi_status: + line.append("OK") + current_table.add_row(line) line = info_iter.next() except (StopIteration, ValueError): # regular end of block @@ -223,8 +235,51 @@ def parse_wmi_table(info, key="Name"): return parsed -def is_wmi_timeout(line): - return line[0] == "WMItimeout" +def _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency): + # Possibilities: + # #1 Agent provides extra column for WMIStatus; since 1.5.0p14 + # <<<SEC>>> + # [foo] + # Name,...,WMIStatus + # ABC,...,OK/Timeout + # [bar] + # Name,...,WMIStatus + # DEF,...,OK/Timeout + # + # #2 Old agents have no WMIStatus column; before 1.5.0p14 + # <<<SEC>>> + # [foo] + # Name,..., + # ABC,..., + # [bar] + # Name,..., + # DEF,..., + # + # #3 Old agents which report a WMITimeout in any sub section; before 1.5.0p14 + # <<<SEC>>> + # [foo] + # WMItimeout + # [bar] + # Name,..., + # DEF,..., + if line[0].lower() == "wmitimeout": + old_timed_out = True + header = ['WMIStatus'] + key = None + else: + old_timed_out = False + header = line[:] + + missing_wmi_status = False + if 'WMIStatus' not in header: + missing_wmi_status = True + header.append('WMIStatus') + + current_table = parsed.setdefault(tablename, + WMITable(tablename, header, key, timestamp, frequency)) + if old_timed_out: + current_table.add_row(['Timeout']) + return missing_wmi_status, current_table #. @@ -255,8 +310,8 @@ def wmi_filter_global_only(tables, row): def _required_tables_missing(tables, required_tables): is_subset = set(required_tables).issubset(set(tables.keys())) - contains_null = any(isinstance(table, NullTable) for table in tables.itervalues()) - return not is_subset or contains_null + contains_timed_out_table = any(table.timed_out for table in tables.itervalues()) + return not is_subset or contains_timed_out_table def inventory_wmi_table_instances(tables, required_tables=None, filt=None, levels=None):

5 years, 3 months

[CMK-1866] - output tables for WMI are modified

by Sergey Kipnis

Module: check_mk Branch: master Commit: 0144523c644a81eec44c1c136fdfcfc51df42949 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=0144523c644a81… Author: Sergey Kipnis <sk(a)mathias-kettner.de> Date: Fri Mar 29 15:56:54 2019 +0100 [CMK-1866] - output tables for WMI are modified - column WMIStatus added with possible content either Timeout or OK - on WMI timeout send cached data - if cached data is absent then send nothing Change-Id: I63a60c90f977425b09f7c1e3715f71496871d41d --- agents/windows/sections/SectionWMI.cc | 102 ++++++++++++++++++++++++++++++++-- agents/windows/sections/SectionWMI.h | 2 + agents/windows/wmiHelper.cc | 16 +++++- 3 files changed, 115 insertions(+), 5 deletions(-) diff --git a/agents/windows/sections/SectionWMI.cc b/agents/windows/sections/SectionWMI.cc index 4c6f1a0..3bace67 100644 --- a/agents/windows/sections/SectionWMI.cc +++ b/agents/windows/sections/SectionWMI.cc @@ -114,7 +114,86 @@ void SectionWMI::suspend(int duration) { _disabled_until = time(nullptr) + duration; } -bool SectionWMI::produceOutputInner(std::ostream &out, +// ******************************************************** +// copy pasted from the new agent, unit-tested in new Agent +// ******************************************************** +namespace cma::tools { +inline std::vector<std::string> SplitString(const std::string &In, + const std::string Delim, + int MaxCount = 0) noexcept { + // sanity + if (In.empty()) return {}; + if (Delim.empty()) return {In}; + + size_t start = 0U; + std::vector<std::string> result; + + auto end = In.find(Delim); + while (end != std::string::npos) { + result.push_back(In.substr(start, end - start)); + + start = end + Delim.length(); + end = In.find(Delim, start); + + // check for a skipping rest + if (result.size() == static_cast<size_t>(MaxCount)) { + end = std::string::npos; + break; + } + } + + auto last_string = In.substr(start, end); + if (!last_string.empty()) result.push_back(last_string); + + return result; +} +} // namespace cma::tools + +// ******************************************************** +// copy pasted from the new agent, unit-tested in new Agent +// ******************************************************** +// adds to the output Table from the WMI WMIStatus column +// column value is either Timeout or OK +// Before +// Name,Freq +// Total,1500 +// AFter +// Name,Freq,WMIStatus +// Total,1500,OK +// Empty or quite short strings are replaced empty string +std::string WmiPostProcess(const std::string &In, bool ExceptionOn, + char Separator) { + if (In.size() < 5) { // 5 is meaningless, just anything low + // data absent + return ExceptionOn ? std::string() : In; + } + + std::string tail_0; + tail_0 += Separator; + tail_0 += "WMIStatus\n"; + + std::string tail_other; + tail_other += Separator; + tail_other += ExceptionOn ? "Timeout\n" : "OK\n"; + + auto table = cma::tools::SplitString(In, "\n"); + size_t s_required = 0; + table[0] += tail_0; + s_required += table[0].size(); + for (size_t i = 1; i < table.size(); ++i) { + table[i] += tail_other; + s_required += table[i].size(); + } + + std::string out; + out.reserve(s_required); + for (const auto line : table) { + out += line; + } + return out; +} + +bool SectionWMI::produceOutputInner(std::ostream &Out, const std::optional<std::string> &) { Debug(_logger) << "SectionWMI::produceOutputInner"; @@ -124,6 +203,8 @@ bool SectionWMI::produceOutputInner(std::ostream &out, bool success = true; + bool exception_on = false; + try { if (_helper.get() == nullptr) { _helper.reset( @@ -150,13 +231,26 @@ bool SectionWMI::produceOutputInner(std::ostream &out, suspend(3600); } + std::stringstream out; outputTable(out, result); + cached_ = out.str(); } catch (const wmi::Timeout &t) { - // Output WMI timeout so that the check in question knows to handle it. - out << t.what() << std::endl; - Debug(_logger) << "SectionWMI::produceOutputInner caught " << t.what(); + exception_on = true; + // only logging + if (cached_.size()) { + Debug(_logger) << "SectionWMI::produceOutputInner caught " + << t.what() << " cached data reused"; + } else { + Debug(_logger) << "SectionWMI::produceOutputInner caught " + << t.what(); + } success = true; } + // in cache we always have last valid data. Or nothing. + // those cached data should be decorated with new column + auto modified = WmiPostProcess(cached_, exception_on, ','); + if (modified.size()) Out << modified; + return success; } diff --git a/agents/windows/sections/SectionWMI.h b/agents/windows/sections/SectionWMI.h index 1eb1b68..801c501 100644 --- a/agents/windows/sections/SectionWMI.h +++ b/agents/windows/sections/SectionWMI.h @@ -57,6 +57,8 @@ private: bool _toggle_if_missing{false}; time_t _disabled_until{0}; std::unique_ptr<wmi::Helper> _helper; + + std::string cached_; // last output stored here, may be reused on timeout }; #endif // SectionWMI_h diff --git a/agents/windows/wmiHelper.cc b/agents/windows/wmiHelper.cc index 482f8b2..fa52685 100644 --- a/agents/windows/wmiHelper.cc +++ b/agents/windows/wmiHelper.cc @@ -221,6 +221,13 @@ vector<wstring> Result::names() const { return result; } +#if 0 +// testing code +// we do not want to loose our time for +// writing real tests for legacy product +bool GlobalTimeout = false; +#endif + bool Result::next() { Debug(_logger) << "Result::next"; if (_enumerator == nullptr) { @@ -229,8 +236,15 @@ bool Result::next() { IWbemClassObject *obj = nullptr; ULONG numReturned = 0; - // always retrieve only one element + HRESULT res = _enumerator->Next(2500, 1, &obj, &numReturned); +#if 0 + // this portion of code is used as embedded testing tool + // looks terrible, still usable + namespace fs = std::filesystem; + std::error_code ec; + if (fs::exists("timeout.txt", ec) || GlobalTimeout) res = WBEM_S_TIMEDOUT; +#endif switch (res) { case WBEM_NO_ERROR:

5 years, 3 months

mk_sap_hana: More generic way to find out SID and INSTANCE

by Simon Betz

Module: check_mk Branch: master Commit: da23bff22c97cf38328aba3eabcc0e2d33656d91 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=da23bff22c97cf… Author: Simon Betz <si(a)mathias-kettner.de> Date: Fri Mar 29 15:30:01 2019 +0100 mk_sap_hana: More generic way to find out SID and INSTANCE Change-Id: I14652079421bf516aab2044da687f627772d967f --- agents/plugins/mk_sap_hana | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/agents/plugins/mk_sap_hana b/agents/plugins/mk_sap_hana index caafd7c..946f1e1 100755 --- a/agents/plugins/mk_sap_hana +++ b/agents/plugins/mk_sap_hana @@ -486,15 +486,15 @@ function do_query () { # | | # '----------------------------------------------------------------------' - # pgrep might not be available on older distros: # shellcheck disable=SC2009 ps -efw | grep "[H]DB.*sapstartsrv" | while read -r line; do - # (!<= is a look behind assertion, so we find \w{3} <- 3 alpha-numerical characters, that follow " /usr/sap/" <- note the leading space - SID=$(echo "$line" | grep -Po '(?<= /usr/sap/)\w{3}') - - # same logic as above, but look for 2 \d digits - INSTANCE=$(echo "$line" | grep -Po "(?<= /usr/sap/${SID}/HDB)\d{2}") + # 'sapstartsrv' may have different base paths: + # /hana/shared/ABC/HDB10/exe/sapstartsrv pf=/hana/shared/ABC/profile/ABC_HDB10_FOOBAR -D -u USER + # /usr/sap/ABC/HDB10/exe/sapstartsrv pf=/hana/shared/ABC/profile/ABC_HDB10_FOOBAR -D -u USER + # From these paths we extract: SID=ABC, INSTANCE=10 + SID=$(echo "$line" | sed 's/.*\/$.*$\/.*\/exe\/sapstartsrv.*/\1/g') + INSTANCE=$(echo "$line" | sed -e 's/.*\/$.*$\/exe\/sapstartsrv.*/\1/g' -e 's/[^0-9]//g') # grep only the first word = the user running the instance INSTANCE_USER=$(echo "$line" | grep -Po "^\s*(\w+)")

5 years, 3 months

Revert "wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable"

by Simon Betz

Module: check_mk Branch: master Commit: 5846b23a6a63d7e4ffa693100f3b5eb19bef234a URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=5846b23a6a63d7… Author: Simon Betz <si(a)mathias-kettner.de> Date: Mon Apr 1 13:48:48 2019 +0200 Revert "wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable" This reverts commit ff4532c4b1e6ef456058205259d2b75ba87c761d. --- checks/wmi.include | 91 ++++++++++++++++-------------------------------------- 1 file changed, 27 insertions(+), 64 deletions(-) diff --git a/checks/wmi.include b/checks/wmi.include index aab6fa6..bbf3420 100644 --- a/checks/wmi.include +++ b/checks/wmi.include @@ -45,6 +45,13 @@ # '----------------------------------------------------------------------' +class NullTable(object): + """Represents an empty table caused by a WMI timeout.""" + + def __getattr__(self, name): + raise MKCounterWrapped('WMI query timed out') + + class WMITable(object): """ Represents a 2-dimensional table of performance metrics. @@ -65,9 +72,10 @@ class WMITable(object): self.__headers = {} self.__timestamp = timestamp self.__frequency = frequency + index = 0 prev_header = None - for index, header in enumerate(headers): + for header in headers: if not header.strip() and prev_header: # MS apparently doesn't bother to provide a name # for base columns with performance counters @@ -75,7 +83,7 @@ class WMITable(object): header = self._normalize_key(header) self.__headers[header] = index prev_header = header - + index += 1 self.__key_index = None if key_field is not None: try: @@ -86,7 +94,6 @@ class WMITable(object): self.__row_lookup = {} self.__rows = [] - self.timed_out = False if rows: for row in rows: self.add_row(row) @@ -124,16 +131,7 @@ class WMITable(object): self.__row_lookup[key] = len(self.__rows) self.__rows.append(row) - if not self.timed_out and self._get_row_col_value(-1, "WMIStatus").lower() == "timeout": - # Check if there's a timeout in the last added line, ie. row (index) == -1 - self.timed_out = True - def get(self, row, column): - if self.timed_out: - raise MKCounterWrapped('WMI query timed out') - return self._get_row_col_value(row, column) - - def _get_row_col_value(self, row, column): if isinstance(row, int): row_index = row else: @@ -200,15 +198,23 @@ def parse_wmi_table(info, key="Name"): # single-table input tablename = "" - missing_wmi_status, current_table =\ - _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency) + if is_wmi_timeout(line): + parsed[tablename] = NullTable() + line = info_iter.next() + continue + + if tablename not in parsed: + parsed[tablename] = WMITable(tablename, line, key, timestamp, frequency) + current = parsed[tablename] # read table content line = info_iter.next() while not line[0].startswith("["): - if missing_wmi_status: - line.append("OK") - current_table.add_row(line) + if is_wmi_timeout(line): + parsed[tablename] = NullTable() + line = info_iter.next() + continue + current.add_row(line) line = info_iter.next() except (StopIteration, ValueError): # regular end of block @@ -217,51 +223,8 @@ def parse_wmi_table(info, key="Name"): return parsed -def _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency): - # Possibilities: - # #1 Agent provides extra column for WMIStatus; since 1.5.0p14 - # <<<SEC>>> - # [foo] - # Name,...,WMIStatus - # ABC,...,OK/Timeout - # [bar] - # Name,...,WMIStatus - # DEF,...,OK/Timeout - # - # #2 Old agents have no WMIStatus column; before 1.5.0p14 - # <<<SEC>>> - # [foo] - # Name,..., - # ABC,..., - # [bar] - # Name,..., - # DEF,..., - # - # #3 Old agents which report a WMITimeout in any sub section; before 1.5.0p14 - # <<<SEC>>> - # [foo] - # WMItimeout - # [bar] - # Name,..., - # DEF,..., - if line[0].lower() == "wmitimeout": - old_timed_out = True - header = ['WMIStatus'] - key = None - else: - old_timed_out = False - header = line[:] - - missing_wmi_status = False - if 'WMIStatus' not in header: - missing_wmi_status = True - header.append('WMIStatus') - - current_table = parsed.setdefault(tablename, - WMITable(tablename, header, key, timestamp, frequency)) - if old_timed_out: - current_table.add_row(['Timeout']) - return missing_wmi_status, current_table +def is_wmi_timeout(line): + return line[0] == "WMItimeout" #. @@ -292,8 +255,8 @@ def wmi_filter_global_only(tables, row): def _required_tables_missing(tables, required_tables): is_subset = set(required_tables).issubset(set(tables.keys())) - contains_timed_out_table = any(table.timed_out for table in tables.itervalues()) - return not is_subset or contains_timed_out_table + contains_null = any(isinstance(table, NullTable) for table in tables.itervalues()) + return not is_subset or contains_null def inventory_wmi_table_instances(tables, required_tables=None, filt=None, levels=None):

5 years, 3 months

wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable

by Simon Betz

Module: check_mk Branch: master Commit: ff4532c4b1e6ef456058205259d2b75ba87c761d URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=ff4532c4b1e6ef… Author: Simon Betz <si(a)mathias-kettner.de> Date: Mon Apr 1 13:29:16 2019 +0200 wmi.include: Refactored WMI timeout handling; Use a more generic way and removed NullTable Change-Id: I81f639d2eed91115340fe3e8e68de08942e021e6 --- checks/wmi.include | 91 ++++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 64 insertions(+), 27 deletions(-) diff --git a/checks/wmi.include b/checks/wmi.include index bbf3420..aab6fa6 100644 --- a/checks/wmi.include +++ b/checks/wmi.include @@ -45,13 +45,6 @@ # '----------------------------------------------------------------------' -class NullTable(object): - """Represents an empty table caused by a WMI timeout.""" - - def __getattr__(self, name): - raise MKCounterWrapped('WMI query timed out') - - class WMITable(object): """ Represents a 2-dimensional table of performance metrics. @@ -72,10 +65,9 @@ class WMITable(object): self.__headers = {} self.__timestamp = timestamp self.__frequency = frequency - index = 0 prev_header = None - for header in headers: + for index, header in enumerate(headers): if not header.strip() and prev_header: # MS apparently doesn't bother to provide a name # for base columns with performance counters @@ -83,7 +75,7 @@ class WMITable(object): header = self._normalize_key(header) self.__headers[header] = index prev_header = header - index += 1 + self.__key_index = None if key_field is not None: try: @@ -94,6 +86,7 @@ class WMITable(object): self.__row_lookup = {} self.__rows = [] + self.timed_out = False if rows: for row in rows: self.add_row(row) @@ -131,7 +124,16 @@ class WMITable(object): self.__row_lookup[key] = len(self.__rows) self.__rows.append(row) + if not self.timed_out and self._get_row_col_value(-1, "WMIStatus").lower() == "timeout": + # Check if there's a timeout in the last added line, ie. row (index) == -1 + self.timed_out = True + def get(self, row, column): + if self.timed_out: + raise MKCounterWrapped('WMI query timed out') + return self._get_row_col_value(row, column) + + def _get_row_col_value(self, row, column): if isinstance(row, int): row_index = row else: @@ -198,23 +200,15 @@ def parse_wmi_table(info, key="Name"): # single-table input tablename = "" - if is_wmi_timeout(line): - parsed[tablename] = NullTable() - line = info_iter.next() - continue - - if tablename not in parsed: - parsed[tablename] = WMITable(tablename, line, key, timestamp, frequency) - current = parsed[tablename] + missing_wmi_status, current_table =\ + _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency) # read table content line = info_iter.next() while not line[0].startswith("["): - if is_wmi_timeout(line): - parsed[tablename] = NullTable() - line = info_iter.next() - continue - current.add_row(line) + if missing_wmi_status: + line.append("OK") + current_table.add_row(line) line = info_iter.next() except (StopIteration, ValueError): # regular end of block @@ -223,8 +217,51 @@ def parse_wmi_table(info, key="Name"): return parsed -def is_wmi_timeout(line): - return line[0] == "WMItimeout" +def _prepare_wmi_table(parsed, tablename, line, key, timestamp, frequency): + # Possibilities: + # #1 Agent provides extra column for WMIStatus; since 1.5.0p14 + # <<<SEC>>> + # [foo] + # Name,...,WMIStatus + # ABC,...,OK/Timeout + # [bar] + # Name,...,WMIStatus + # DEF,...,OK/Timeout + # + # #2 Old agents have no WMIStatus column; before 1.5.0p14 + # <<<SEC>>> + # [foo] + # Name,..., + # ABC,..., + # [bar] + # Name,..., + # DEF,..., + # + # #3 Old agents which report a WMITimeout in any sub section; before 1.5.0p14 + # <<<SEC>>> + # [foo] + # WMItimeout + # [bar] + # Name,..., + # DEF,..., + if line[0].lower() == "wmitimeout": + old_timed_out = True + header = ['WMIStatus'] + key = None + else: + old_timed_out = False + header = line[:] + + missing_wmi_status = False + if 'WMIStatus' not in header: + missing_wmi_status = True + header.append('WMIStatus') + + current_table = parsed.setdefault(tablename, + WMITable(tablename, header, key, timestamp, frequency)) + if old_timed_out: + current_table.add_row(['Timeout']) + return missing_wmi_status, current_table #. @@ -255,8 +292,8 @@ def wmi_filter_global_only(tables, row): def _required_tables_missing(tables, required_tables): is_subset = set(required_tables).issubset(set(tables.keys())) - contains_null = any(isinstance(table, NullTable) for table in tables.itervalues()) - return not is_subset or contains_null + contains_timed_out_table = any(table.timed_out for table in tables.itervalues()) + return not is_subset or contains_timed_out_table def inventory_wmi_table_instances(tables, required_tables=None, filt=None, levels=None):

5 years, 3 months

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

Checkmk git commits April 2019