log update

hstreamdb · Jun 28, 2024 · d0ec907 · d0ec907
1 parent f8f9c9f
commit d0ec907
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 16 deletions.
diff --git a/hstream-store/cbits/logdevice/ld_health_checker.cpp b/hstream-store/cbits/logdevice/ld_health_checker.cpp
@@ -45,7 +45,7 @@ class LdChecker {
     auto duration =
         std::chrono::duration_cast<std::chrono::milliseconds>(end - start)
             .count();
-    ld_warning("GetClusterState took %ld ms", duration);
+    ld_debug("GetClusterState took %ld ms", duration);
 
     // getClusterState(*client_impl, *nodes_configuration);
 
@@ -69,7 +69,7 @@ class LdChecker {
     if (!unhealthy_nodes_set.empty()) {
       ld_warning("Cluster has %lu unhealthy nodes:",
                  unhealthy_nodes_set.size());
-      // printUnhealthyNodes(*nodes_configuration, unhealthy_nodes_set);
+      printUnhealthyNodes(*nodes_configuration, unhealthy_nodes_set);
     }
 
     return unhealthy_nodes_set.size() <= unhealthy_node_limit;
@@ -103,7 +103,7 @@ class LdChecker {
                  res.node_id, res.addr.c_str(), st, deadNodes.c_str(),
                  unhealthyNodes.c_str());
     }
-    ld_warning("Check return unhealthy nodes: [%s]",
+    ld_warning("Check return unhealthy nodes index: [%s]",
                folly::join(',', sets).c_str());
   }
 

diff --git a/hstream/src/HStream/Server/CacheStore.hs b/hstream/src/HStream/Server/CacheStore.hs
@@ -313,7 +313,7 @@ appendHStoreWithRetry ldClient shardId payload cmpStrategy dumpState = do
                loop cnt'
              _ -> loop exitNum
      | cnt == exitNum = do
-       Log.warning $ "Dump to shardId " <> Log.build shardId <> " failed because cache store is not dumping, will retry later."
+       Log.warning $ "Dump to shardId " <> Log.build shardId <> " failed because cache store is not in dumping state, will retry later."
        return Nothing
      | otherwise = do
        Log.fatal $ "Dump to shardId " <> Log.build shardId <> " failed after exausting the retry attempts, drop the record."

diff --git a/hstream/src/HStream/Server/HealthMonitor.hs b/hstream/src/HStream/Server/HealthMonitor.hs
@@ -40,7 +40,7 @@ startMonitor :: ServerContext -> HealthMonitor -> Int -> IO ()
 startMonitor sc hm delaySecond = forever $ do
   threadDelay $ delaySecond * 1000 * 1000
   start <- getCurrentTime
-  Log.debug $ "========== docheck start..." <> " in " <> Log.build (show start)
+  -- Log.debug $ "========== docheck start..." <> " in " <> Log.build (show start)
   res <- try @SomeException $ docheck sc hm
   end <- getCurrentTime
   case res of
@@ -49,9 +49,9 @@ startMonitor sc hm delaySecond = forever $ do
   let diff = nominalDiffTimeToSeconds $ diffUTCTime end start
   when (diff > 1) $
     Log.warning $ "Monitor check return slow, total use " <> Log.build (show diff) <> "s"
-  Log.debug $ "========== docheck end..." <> " in " <> Log.build (show end)
-          <> ", with start time: " <> Log.build (show start)
-          <> ", duration: " <> Log.build (show diff)
+  Log.debug $ "Health monitor finish check in " <> Log.build (show end)
+           <> ", with start time: " <> Log.build (show start)
+           <> ", duration: " <> Log.build (show diff)
 
 docheck :: ServerContext -> HealthMonitor -> IO ()
 docheck sc@ServerContext{..} hm = do
@@ -80,21 +80,19 @@ checkLdCluster HealthMonitor{..} = do
   start <- getTime Monotonic
   res <- S.isLdClusterHealthy ldChecker ldUnhealthyNodesLimit
   end <- getTime Monotonic
-  let sDuration = toNanoSecs (diffTimeSpec end start) `div` 1000000
-  if sDuration > 1000
-    then Log.warning $ "CheckLdCluster slow, total time " <> Log.build sDuration <> "ms"
-    else Log.debug $ "Finish checkLdClusster, total time " <> Log.build sDuration <> "ms"
+  let msDuration = toNanoSecs (diffTimeSpec end start) `div` 1000000
+  when (msDuration > 1000) $
+    Log.warning $ "CheckLdCluster return slow, total time " <> Log.build msDuration <> "ms"
   return res
 
 checkMeta :: HealthMonitor -> IO Bool
 checkMeta HealthMonitor{..} | ZKHandle c <- metaHandle = do
   start <- getTime Monotonic
   res <- checkRecoverable =<< unsafeGetZHandle c
   end <- getTime Monotonic
-  let sDuration = toNanoSecs (diffTimeSpec end start) `div` 1000000
-  if sDuration > 1000
-    then Log.warning $ "CheckMeta slow, total time " <> Log.build sDuration <> "ms"
-    else Log.debug $ "Finish checkMeta, total time " <> Log.build sDuration <> "ms"
+  let msDuration = toNanoSecs (diffTimeSpec end start) `div` 1000000
+  when (msDuration > 1000) $
+    Log.warning $ "CheckMeta return slow, total time " <> Log.build msDuration <> "ms"
   return res
 checkMeta HealthMonitor{..} | _ <- metaHandle = do
   return True