Module: check_mk Branch: master Commit: 110cd6b42743bc5470cb32f8fa2baf5ab6c40367 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=110cd6b42743bc… Author: Lars Michelsen <lm(a)mathias-kettner.de> Date: Tue Nov 25 10:32:32 2014 +0100 #1532 FIX mk_sap: Cleaning up old state information from sap.state file The mk_sap agent plugin writes a state file which contains information for each SAP alert log file which is monitored. Old log files are now cleaned up from this file, when they are not monitored anymore. --- .werks/1532 | 11 +++++++++++ ChangeLog | 1 + agents/plugins/mk_sap | 46 +++++++++++++++++++++++++++++++++++++--------- 3 files changed, 49 insertions(+), 9 deletions(-) diff --git a/.werks/1532 b/.werks/1532 new file mode 100644 index 0000000..bf6feb4 --- /dev/null +++ b/.werks/1532 @@ -0,0 +1,11 @@ +Title: mk_sap: Cleaning up old state information from sap.state file +Level: 1 +Component: checks +Compatible: compat +Version: 1.2.5i7 +Date: 1416907854 +Class: fix + +The mk_sap agent plugin writes a state file which contains information +for each SAP alert log file which is monitored. Old log files are now +cleaned up from this file, when they are not monitored anymore. diff --git a/ChangeLog b/ChangeLog index a5092a2..bc3f9df 100644 --- a/ChangeLog +++ b/ChangeLog @@ -38,6 +38,7 @@ * 1503 FIX: Fixed monitoring of multiple SAP instances with one mk_sap plugin... * 1515 FIX: cisco_secure: fix service description, fix OK state in case of no violation * 1449 FIX: nginx_status: agent plugin no longer honours "http(s)_proxy" env variables of root user + * 1532 FIX: mk_sap: Cleaning up old state information from sap.state file... Multisite: * 1508 Allow input of plugin output and perfdata when faking check results... diff --git a/agents/plugins/mk_sap b/agents/plugins/mk_sap index 8f867df..747f9ae 100755 --- a/agents/plugins/mk_sap +++ b/agents/plugins/mk_sap @@ -109,6 +109,7 @@ SKIP_MTCLASSES = [ ] STATE_FILE = '/etc/check_mk/sap.state' +state_file_changed = False # ############################################################################# @@ -139,7 +140,6 @@ monitor_paths = [ # 'SAP CCMS Monitor Templates/Dialog Overview/*', ] monitor_types = [] -# henri config_file = os.getenv('MK_CONFDIR', '/etc/check_mk') + '/sap.cfg' cfg = {} @@ -156,6 +156,10 @@ try: except IOError: states = {} +# index of all logfiles which have been found in a run. This is used to +# remove logfiles which are not available anymore from the states dict. +logfiles = [] + # ############################################################################# # @@ -322,6 +326,8 @@ def alert_details(cfg, alert): return state, msg def process_alerts(cfg, logs, ms_name, mon_name, node, alerts): + global state_file_changed + sid = node["MTSYSID"].strip() or 'Other' context = node["MTMCNAME"].strip() or 'Other' path = node["PATH"] @@ -330,6 +336,8 @@ def process_alerts(cfg, logs, ms_name, mon_name, node, alerts): hostname = sid logfile = context + "/" + path + logfiles.append((hostname, logfile)) + logs.setdefault(sid, {}) logs[hostname][logfile] = [] newest_log_dt = None @@ -350,7 +358,7 @@ def process_alerts(cfg, logs, ms_name, mon_name, node, alerts): if newest_log_dt: # Write newest log age to cache to prevent double processing of logs states[(hostname, logfile)] = newest_log_dt - file(STATE_FILE, 'w').write(repr(states)) + state_file_changed = True return logs @@ -452,12 +460,32 @@ def check(cfg): # It is possible to configure multiple SAP instances to monitor. Loop them all, but # do not terminate when one connection failed -for entry in cfg: - try: - check(entry) - except sapnwrfc.RFCCommunicationError, e: - sys.stderr.write('ERROR: Unable to connect (%s)\n' % e) - except: - sys.stderr.write('ERROR: Unhandled exception (%s)\n' % e) +processed_all = True +try: + for entry in cfg: + try: + check(entry) + except sapnwrfc.RFCCommunicationError, e: + sys.stderr.write('ERROR: Unable to connect (%s)\n' % e) + processed_all = False + except Exception, e: + sys.stderr.write('ERROR: Unhandled exception (%s)\n' % e) + processed_all = False + + # Now check whether or not an old logfile needs to be removed. This can only + # be done this way, when all hosts have been reached. Otherwise the cleanup + # is skipped. + if processed_all: + for key in states.keys(): + if key not in logfiles: + state_file_changed = True + del states[key] + + # Only write the state file once per run. And only when it has been changed + if state_file_changed: + file(STATE_FILE, 'w').write(repr(states)) + +except Exception, e: + sys.stderr.write('ERROR: Unhandled exception (%s)\n' % e) sys.exit(0)