X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=policy.py;h=cdd311c043fd2547c465b7ff10577a419f78ff74;hb=0597380e64582380753244a49b5f10e90049eae9;hp=edff7bdd279b10fdb6708bee188416f33ef7a920;hpb=4cb581d60be6b121a387f5ff2d033fa17e2a6d3a;p=monitor.git diff --git a/policy.py b/policy.py index edff7bd..cdd311c 100755 --- a/policy.py +++ b/policy.py @@ -27,7 +27,7 @@ from monitor.wrapper import plccache from monitor.database.info.model import * from monitor.database.info.interface import * -from nodequery import verify,query_to_dict,node_select +from monitor.query import verify,query_to_dict,node_select api = plc.getAuthAPI() @@ -76,7 +76,7 @@ def main(hostnames, sitenames): # NOTE: there is a narrow window in which this command must be # evaluated, otherwise the notice will not go out. # this is not ideal. - sitehist.sendMessage('online_notice', hostname=host, viart=False) + sitehist.sendMessage('online_notice', hostname=host, viart=False, saveact=True) print "send message for host %s online" % host @@ -91,25 +91,62 @@ def main(hostnames, sitenames): # if it is offline and HAS a PCU, then try to use it. if nodehist.haspcu and nodehist.status in ['offline', 'down'] and \ changed_greaterthan(nodehist.last_changed,1.0) and \ + not nodehist.firewall and \ not found_between(recent_actions, 'try_reboot', 3.5, 1): + # TODO: there MUST be a better way to do this... + # get fb node record for pcuid + fbpcu = None + fbnode = FindbadNodeRecord.get_latest_by(hostname=host) + if fbnode: + fbpcu = FindbadPCURecord.get_latest_by(plc_pcuid=fbnode.plc_pcuid) + sitehist.attemptReboot(host) print "send message for host %s try_reboot" % host + if not fbpcu.test_is_ok() and \ + not found_within(recent_actions, 'pcuerror_notice', 3.0): + + args = {} + if fbpcu: + args['pcu_name'] = fbpcu.pcu_name() + args['pcu_errors'] = fbpcu.pcu_errors() + args['plc_pcuid'] = fbpcu.plc_pcuid + else: + args['pcu_name'] = "error looking up pcu name" + args['pcu_errors'] = "" + args['plc_pcuid'] = 0 + + args['hostname'] = host + sitehist.sendMessage('pcuerror_notice', **args) + print "send message for host %s PCU Failure" % host + # NOTE: non-intuitive is that found_between(try_reboot, 3.5, 1) # will be false for a day after the above condition is satisfied if nodehist.haspcu and nodehist.status in ['offline', 'down'] and \ changed_greaterthan(nodehist.last_changed,1.5) and \ + not nodehist.firewall and \ found_between(recent_actions, 'try_reboot', 3.5, 1) and \ not found_within(recent_actions, 'pcufailed_notice', 3.5): + # TODO: there MUST be a better way to do this... + # get fb node record for pcuid + fbpcu = None + fbnode = FindbadNodeRecord.get_latest_by(hostname=host) + if fbnode: + fbpcu = FindbadPCURecord.get_latest_by(plc_pcuid=fbnode.plc_pcuid) + if fbpcu: + pcu_name = fbpcu.pcu_name() + else: + pcu_name = "error looking up pcu name" + + # get fb pcu record for pcuid # send pcu failure message - #act = ActionRecord(**kwargs) - sitehist.sendMessage('pcufailed_notice', hostname=host) + sitehist.sendMessage('pcufailed_notice', hostname=host, pcu_name=pcu_name) print "send message for host %s PCU Failure" % host - if nodehist.status == 'monitordebug' and \ - changed_greaterthan(nodehist.last_changed, 1) and \ + if nodehist.status == 'failboot' and \ + changed_greaterthan(nodehist.last_changed, 0.25) and \ not found_between(recent_actions, 'bootmanager_restore', 0.5, 0): # send down node notice # delay 0.5 days before retrying... @@ -119,12 +156,17 @@ def main(hostnames, sitenames): # sitehist.sendMessage('retry_bootman', hostname=host) if nodehist.status == 'down' and \ - changed_greaterthan(nodehist.last_changed, 2) and \ - not found_within(recent_actions, 'down_notice', 3.5): - # send down node notice - - sitehist.sendMessage('down_notice', hostname=host) - print "send message for host %s down" % host + changed_greaterthan(nodehist.last_changed, 2): + if not nodehist.firewall and not found_within(recent_actions, 'down_notice', 3.5): + # send down node notice + sitehist.sendMessage('down_notice', hostname=host) + print "send message for host %s down" % host + + if nodehist.firewall and not found_within(recent_actions, 'firewall_notice', 3.5): + # send down node notice + #email_exception(host, "firewall_notice") + sitehist.sendMessage('firewall_notice', hostname=host) + print "send message for host %s down" % host node_count = node_count + 1 print "time: ", time.strftime('%Y-%m-%d %H:%M:%S') @@ -134,9 +176,13 @@ def main(hostnames, sitenames): for i,site in enumerate(sitenames): sitehist = SiteInterface.get_or_make(loginbase=site) siteblack = BlacklistRecord.get_by(loginbase=site) + skip_due_to_blacklist=False if siteblack and not siteblack.expired(): print "skipping %s due to blacklist. will expire %s" % (site, siteblack.willExpire() ) + skip_due_to_blacklist=True + sitehist.clearPenalty() + sitehist.applyPenalty() continue # TODO: make query only return records within a certin time range, @@ -145,40 +191,46 @@ def main(hostnames, sitenames): print "%s %s %s" % (i, sitehist.db.loginbase, sitehist.db.status) - # determine if there are penalties within the last 30 days? - # if so, add a 'pause_penalty' action. - if sitehist.db.message_id != 0 and sitehist.db.message_status == 'open' and \ - sitehist.db.penalty_level > 0 and not found_within(recent_actions, 'pause_penalty', 30): - # pause escalation - print "Pausing penalties for %s" % site - sitehist.pausePenalty() - else: + if sitehist.db.status == 'down': + if sitehist.db.penalty_pause and \ + changed_greaterthan(sitehist.db.penalty_pause_time, 30): + + email_exception("", "clear pause penalty for site: %s" % sitehist.db.loginbase) + sitehist.closeTicket() + # NOTE: but preserve the penalty status. + sitehist.clearPenaltyPause() + + if sitehist.db.message_id != 0 and \ + sitehist.db.message_status == 'open' and \ + not sitehist.db.penalty_pause: + + email_exception("", "pause penalty for site: %s" % sitehist.db.loginbase) + sitehist.setPenaltyPause() - if sitehist.db.status == 'down': - if not found_within(recent_actions, 'pause_penalty', 30) and \ - not found_within(recent_actions, 'increase_penalty', 7) and \ - changed_greaterthan(sitehist.db.last_changed, 7): + if not sitehist.db.penalty_pause and \ + not found_within(recent_actions, 'increase_penalty', 7) and \ + changed_greaterthan(sitehist.db.last_changed, 7): - # TODO: catch errors - sitehist.increasePenalty() - sitehist.applyPenalty() - sitehist.sendMessage('increase_penalty') + # TODO: catch errors + sitehist.increasePenalty() + sitehist.applyPenalty() + sitehist.sendMessage('increase_penalty') - print "send message for site %s penalty increase" % site + print "send message for site %s penalty increase" % site - if sitehist.db.status == 'good': - # clear penalty - # NOTE: because 'all clear' should have an indefinite status, we - # have a boolean value rather than a 'recent action' - if sitehist.db.penalty_applied: - # send message that penalties are cleared. + if sitehist.db.status == 'good': + # clear penalty + # NOTE: because 'all clear' should have an indefinite status, we + # have a boolean value rather than a 'recent action' + if sitehist.db.penalty_applied or sitehist.db.penalty_pause: + # send message that penalties are cleared. - sitehist.clearPenalty() - sitehist.applyPenalty() - sitehist.sendMessage('clear_penalty') - sitehist.closeTicket() + sitehist.clearPenalty() + sitehist.applyPenalty() + sitehist.sendMessage('clear_penalty') + sitehist.closeTicket() - print "send message for site %s penalty cleared" % site + print "send message for site %s penalty cleared" % site site_count = site_count + 1 @@ -253,6 +305,6 @@ if __name__ == "__main__": session.flush() sys.exit(0) except: - #email_exception() + email_exception() print traceback.print_exc(); print "fail all..."