Fix rtt, including bug in time compute

hbd/server/udp.py

@@ -68,7 +68,6 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
       - config: dict of configuration
       - hbdclass: module providing Host/Connection classes
       - log: callable(loghost, message)
-      - pushmsg: callable(message)
       - msg_to_websockets: callable(typ, data)
       - msg_journal: MessageJournal instance for logging all messages
       - DEBUG, verbose
@@ -91,7 +90,6 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
     cfg = ctx.get("config", {})
     hbdcls = ctx.get("hbdclass")
     log = ctx.get("log")
-    pushmsg = ctx.get("pushmsg")
     msg_to_websockets = ctx.get("msg_to_websockets")
     DEBUG = ctx.get("DEBUG", 0)
     verbose = ctx.get("verbose", False)
@@ -100,18 +98,24 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
     ip = addr[0] if isinstance(addr, (list, tuple)) else addr
     name = msg.get("name", "unknown")
     from ..common.utils import shortname
+    from . import config as config_mod
 
     uname = shortname(name)
 
     if uname not in hbdcls.Host.hosts:
         host = hbdcls.Host(uname)
-        host.dyn = uname in cfg.get("dyndnshosts", [])
+        # Use new config function to check dyndns
+        dyndnshosts = config_mod.get_dyndnshosts(cfg)
+        host.dyn = uname in dyndnshosts
         if verbose:
             print(("XX: New host, num now %s" % (len(hbdcls.Host.hosts))))
         newh = True
     else:
         host = hbdcls.Host.hosts[uname]
         newh = False
+    
+    # Get watchhosts once for use throughout message handling
+    watchhosts = config_mod.get_watchhosts(cfg)
 
     cid = msg.get("id", 0)
     try:
@@ -181,9 +185,8 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
 
     if res:
         eventlog(uname, "WARNING", res)
-        if uname in cfg.get("watchhosts", []):
-            if pushmsg:
-                pushmsg("%s %s" % (host.name, res))
+        if uname in watchhosts:
+            notify_mod.pushmsg_for_host(uname, "%s %s" % (host.name, res))
 
     interval = int(msg.get("interval", 0) or 0)
     shutdown = msg.get("shutdown", 0)
@@ -193,15 +196,13 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
 
     if boot:
         eventlog(uname, "INFO", "booted")
-        if uname in cfg.get("watchhosts", []):
+        if uname in watchhosts:
             m = "%s booted" % (host.name)
-            if pushmsg:
-                pushmsg(m)
+            notify_mod.pushmsg_for_host(uname, m)
     if message:
         eventlog(uname, "INFO", "msg: %s" % message, service=service)
-        if uname in cfg.get("watchhosts", []):
-            if pushmsg:
-                pushmsg(message)
+        if uname in watchhosts:
+            notify_mod.pushmsg_for_host(uname, message)
 
     if conn.getstate() != hbdcls.Connection.UP:
         lasts = conn.state
@@ -211,9 +212,8 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
         else:
             m = "%s back after being %s for %s" % (conn.afam, lasts, dur(d))
         eventlog(uname, "RECOVER", m)
-        if uname in cfg.get("watchhosts", []):
-            if pushmsg:
-                pushmsg("%s %s is back" % (uname, conn.afam))
+        if uname in watchhosts:
+            notify_mod.pushmsg_for_host(uname, "%s %s is back" % (uname, conn.afam))
 
     if boot or newh:
         host.upcount = host.doesack
@@ -222,9 +222,8 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
 
     if shutdown:
         eventlog(uname, "INFO", "%s shutdown" % conn.afam)
-        if uname in cfg.get("watchhosts", []):
-            if pushmsg:
-                pushmsg("%s %s shutdown" % (uname, conn.afam))
+        if uname in watchhosts:
+            notify_mod.pushmsg_for_host(uname, "%s %s shutdown" % (uname, conn.afam))
         conn.newstate(hbdcls.Connection.DOWN, now)
 
     if interval > 0:
@@ -247,11 +246,21 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
                 connection.newstate(hbdcls.Connection.OVERDUE, now, cfg.get("grace", 2))
                 
                 msg = f"{connection.afam} overdue"
-                eventlog(uname, "CRITICAL" if uname in cfg.get("watchhosts", []) else "WARNING", msg)
+                eventlog(uname, "CRITICAL" if uname in watchhosts else "WARNING", msg)
                 
-                if uname in cfg.get("watchhosts", []):
-                    if pushmsg:
-                        pushmsg(f"{uname} {msg}")
+                if uname in watchhosts:
+                    notify_mod.pushmsg_for_host(uname, f"{uname} {msg}")
+                
+                # Check RTT thresholds with infinite RTT for overdue hosts
+                threshold_checker = ctx.get("threshold_checker")
+                if threshold_checker:
+                    metric_path = "rtt"
+                    threshold_checker.check_value(
+                        host_name=uname,
+                        metric_path=metric_path,
+                        value=float('inf'),
+                        alert_states=host.alert_states
+                    )
                 
                 # Notify websockets
                 if msg_to_websockets:
@@ -274,8 +283,8 @@ def handle_datagram(msg: dict, addr, transport, ctx: dict):
     # Check RTT thresholds using the threshold checker
     threshold_checker = ctx.get("threshold_checker")
     if threshold_checker and rtt and rtt > 0:
-        # Metric path for RTT is "rtt.<hostname>"
-        metric_path = f"rtt.{uname}"
+        # Metric path for RTT is simply "rtt"
+        metric_path = "rtt"
         
         # Check against configured thresholds (handles alerts, notifications, etc.)
         threshold_checker.check_value(