merge from 2.0 branch
[monitor.git] / policy.py
index 4befbd9..43b37ca 100755 (executable)
--- a/policy.py
+++ b/policy.py
@@ -47,6 +47,7 @@ def main(hostnames, sitenames):
                        lb = plccache.plcdb_hn2lb[host]
                except:
                        print "unknown host in plcdb_hn2lb %s" % host
+                       email_exception(host)
                        continue
 
                nodeblack = BlacklistRecord.get_by(hostname=host)
@@ -64,35 +65,46 @@ def main(hostnames, sitenames):
                print "%s %s %s" % (i, nodehist.hostname, nodehist.status)
                if nodehist.status == 'good' and \
                        changed_lessthan(nodehist.last_changed, 1.0) and \
+                       found_within(recent_actions, 'down_notice', 7.0) and \
                        not found_within(recent_actions, 'online_notice', 0.5):
+                           # NOTE: searching for down_notice proves that the node has
+                               #               gone through a 'down' state first, rather than just
+                               #               flapping through: good, offline, online, ...
+                               #       
                                # NOTE: there is a narrow window in which this command must be
-                               # evaluated, otherwise the notice will not go out.  this is not ideal.
+                               #               evaluated, otherwise the notice will not go out.  
+                               #               this is not ideal.
                                sitehist.sendMessage('online_notice', hostname=host, viart=False)
                                print "send message for host %s online" % host
 
-                               pass
 
-               if ( nodehist.status == 'offline' or nodehist.status == 'down' ) and \
+               # if a node is offline and doesn't have a PCU, remind the user that they should have one.
+               if not nodehist.haspcu and nodehist.status in ['offline', 'down'] and \
                        changed_greaterthan(nodehist.last_changed,1.0) and \
-                       not found_between(recent_actions, 'first_try_reboot', 3.5, 1):
+                       not found_within(recent_actions, 'pcumissing_notice', 7.0):
+
+                               sitehist.sendMessage('pcumissing_notice', hostname=host)
+                               print "send message for host %s pcumissing_notice" % host
+
+               # 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 found_between(recent_actions, 'try_reboot', 3.5, 1):
 
                                sitehist.attemptReboot(host)
-                               print "send message for host %s first_try_reboot" % host
-                               pass
+                               print "send message for host %s try_reboot" % host
 
-               # NOTE: non-intuitive is that found_between(first_try_reboot, 3.5, 1)
+               # 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.status == 'offline' or nodehist.status == 'down' ) and \
+               if nodehist.haspcu and nodehist.status in ['offline', 'down'] and \
                        changed_greaterthan(nodehist.last_changed,1.5) and \
-                       found_between(recent_actions, 'first_try_reboot', 3.5, 1) and \
+                       found_between(recent_actions, 'try_reboot', 3.5, 1) and \
                        not found_within(recent_actions, 'pcufailed_notice', 3.5):
-                       # found_within(recent_actions, 'first_try_reboot', 3.5) and \
                                
                                # send pcu failure message
                                #act = ActionRecord(**kwargs)
                                sitehist.sendMessage('pcufailed_notice', hostname=host)
                                print "send message for host %s PCU Failure" % host
-                               pass
 
                if nodehist.status == 'monitordebug' and \
                        changed_greaterthan(nodehist.last_changed, 1) and \
@@ -111,9 +123,10 @@ def main(hostnames, sitenames):
 
                                sitehist.sendMessage('down_notice', hostname=host)
                                print "send message for host %s down" % host
-                               pass
 
                node_count = node_count + 1
+               print "time: ", time.strftime('%Y-%m-%d %H:%M:%S')
+               sys.stdout.flush()
                session.flush()
 
        for i,site in enumerate(sitenames):
@@ -158,13 +171,16 @@ def main(hostnames, sitenames):
                # find all ticket ids for site ( could be on the site record? )
                # 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:
+               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()
 
                site_count = site_count + 1
 
+               print "time: ", time.strftime('%Y-%m-%d %H:%M:%S')
+               sys.stdout.flush()
                session.flush()
 
        session.flush()
@@ -227,6 +243,7 @@ if __name__ == "__main__":
 
        try:
                main(hostnames, sitenames)
+               session.flush()
        except KeyboardInterrupt:
                print "Killed by interrupt"
                session.flush()