From: Jeffrey Altman <jaltman@secure-endpoints.com>
Date: Mon, 11 May 2009 15:16:32 +0000 (+0000)
Subject: DEVEL15-windows-hard-and-idle-dead-timeouts-20090511
X-Git-Tag: openafs-devel-1_5_60~70
X-Git-Url: https://git.michaelhowe.org/gitweb/?a=commitdiff_plain;h=1c7cc2997122d02a6e76fc6f38511c41864642e4;p=packages%2Fo%2Fopenafs.git

DEVEL15-windows-hard-and-idle-dead-timeouts-20090511

LICENSE MIT

Fix conditional test that assigns value to idleDeadTimeout

Modify event log message to indicate timeout could be either a
hard or an idle timeout.

As described in the source comments, do not retry on a hard or
idle timeout.


(cherry picked from commit f5a97a145e09a16a9b2d3ab485f33f520235b7e9)
---

diff --git a/src/WINNT/afsd/afsd_eventmessages.mc b/src/WINNT/afsd/afsd_eventmessages.mc
index 53f48a618..37c76a0be 100644
--- a/src/WINNT/afsd/afsd_eventmessages.mc
+++ b/src/WINNT/afsd/afsd_eventmessages.mc
@@ -163,7 +163,7 @@ Severity=Warning
 Facility=System
 SymbolicName=MSG_RX_HARD_DEAD_TIME_EXCEEDED
 Language=English
-Hard Dead Time exceeded when communicating with server %1
+Hard or Idle Dead Time exceeded when communicating with server %1
 .
 
 MessageId=
diff --git a/src/WINNT/afsd/cm_conn.c b/src/WINNT/afsd/cm_conn.c
index cabdf9029..cc70fb93f 100644
--- a/src/WINNT/afsd/cm_conn.c
+++ b/src/WINNT/afsd/cm_conn.c
@@ -107,7 +107,7 @@ void cm_InitConn(void)
 	    HardDeadtimeout = (unsigned short) RDRtimeout;
             afsi_log("HardDeadTimeout is %d", HardDeadtimeout);
         }
-	if (ConnDeadtimeout == 0) {
+	if (IdleDeadtimeout == 0) {
 	    IdleDeadtimeout = (unsigned short) RDRtimeout;
             afsi_log("IdleDeadTimeout is %d", IdleDeadtimeout);
         }
@@ -604,7 +604,8 @@ cm_Analyze(cm_conn_t *connp, cm_user_t *userp, cm_req_t *reqp,
 
     /* RX codes */
     else if (errorCode == RX_CALL_TIMEOUT) {
-        /* server took longer than hardDeadTime 
+        /* RPC took longer than hardDeadTime or the server
+         * reported idle for longer than idleDeadTime
          * don't mark server as down but don't retry
          * this is to prevent the SMB session from timing out
          * In addition, we log an event to the event log 
@@ -620,11 +621,10 @@ cm_Analyze(cm_conn_t *connp, cm_user_t *userp, cm_req_t *reqp,
 
             LogEvent(EVENTLOG_WARNING_TYPE, MSG_RX_HARD_DEAD_TIME_EXCEEDED, addr);
 	  
-            osi_Log1(afsd_logp, "cm_Analyze: hardDeadTime exceeded addr[%s]",
+            osi_Log1(afsd_logp, "cm_Analyze: hardDeadTime or idleDeadTime exceeded addr[%s]",
                      osi_LogSaveString(afsd_logp,addr));
             reqp->tokenIdleErrorServp = serverp;
             reqp->idleError++;
-            retry = 1;
         }
     }
     else if (errorCode >= -64 && errorCode < 0) {