From f74c293dd2e6b80a8e6011628e71a00547a91611 Mon Sep 17 00:00:00 2001
From: David Anderson <davea@ssl.berkeley.edu>
Date: Sun, 12 Mar 2006 06:20:28 +0000
Subject: [PATCH] scheduler fix

svn path=/trunk/boinc/; revision=9633
---
 checkin_notes          | 14 ++++++++++++++
 sched/handle_request.C | 33 +++++++++++++++++++++++++++++++--
 sched/transitioner.C   |  2 +-
 3 files changed, 46 insertions(+), 3 deletions(-)

diff --git a/checkin_notes b/checkin_notes
index 1f96ac41ec..56f6568adf 100755
--- a/checkin_notes
+++ b/checkin_notes
@@ -2773,3 +2773,17 @@ Rom    11 Mar 2006
     api/
         boinc_api.C, .h
         windows_opengl.C
+
+David  11 Mar 2006
+    - Scheduler (see checkin of 9 Mar): When we use a host record
+        based on host CPID (presumably because user detached/reattached)
+        mark all in-progress results as over, client error.
+        This prevents them from being resent.
+        Avoid a situation where a result causes problems on a host,
+        user detaches/reattaches to clear things up,
+        and then the host gets sent the same result again.
+        (from Bruce Allen)
+
+    sched/
+        handle_request.C
+        transitioner.C
diff --git a/sched/handle_request.C b/sched/handle_request.C
index 46eff8a2e2..e6e0059018 100644
--- a/sched/handle_request.C
+++ b/sched/handle_request.C
@@ -108,7 +108,7 @@ void unlock_sched(SCHEDULER_REPLY& reply) {
     close(reply.lockfile_fd);
 }
 
-// find the user's most recent host with given host CPID
+// find the user's most recently-created host with given host CPID
 //
 bool find_host_by_cpid(DB_USER& user, char* host_cpid, DB_HOST& host) {
     char buf[256], buf2[256];
@@ -125,6 +125,33 @@ bool find_host_by_cpid(DB_USER& user, char* host_cpid, DB_HOST& host) {
     return false;
 }
 
+// scan in-progress results for the given host,
+// and mark them as done, client error
+//
+void mark_results_aborted(DB_HOST& host) {
+    char buf[256], buf2[256];
+    DB_RESULT result;
+    sprintf(buf, "where hostid=%d and server_state=%d",
+        host.id,
+        RESULT_SERVER_STATE_IN_PROGRESS
+    );
+    while (!result.enumerate(buf)) {
+        sprintf(buf2,
+            "server_state=%d, outcome=%d",
+            RESULT_SERVER_STATE_OVER,
+            RESULT_OUTCOME_CLIENT_ERROR
+        );
+        result.update_field(buf2);
+
+        // and trigger WU transition
+        //
+        DB_WORKUNIT wu;
+        wu.id = result.workunitid;
+        sprintf(buf2, "transition_time=%d", time(0));
+        wu.update_field(buf2);
+    }
+}
+
 // Based on the info in the request message,
 // look up the host and its user, and make sure the authenticator matches.
 // Some special cases:
@@ -262,10 +289,12 @@ lookup_user_and_make_new_host:
         // scan backwards through this user's hosts,
         // looking for one with the same host CPID.
         // If we find one, it means the user detached and reattached.
-        // Use the existing host record.
+        // Use the existing host record,
+        // and mark in-progress results as aborted.
         //
         if (strlen(sreq.host.host_cpid)) {
             if (find_host_by_cpid(user, sreq.host.host_cpid, host)) {
+                mark_results_aborted(host);
                 goto got_host;
             }
         }
diff --git a/sched/transitioner.C b/sched/transitioner.C
index 7204be7e61..40cfd815d4 100644
--- a/sched/transitioner.C
+++ b/sched/transitioner.C
@@ -66,7 +66,7 @@ int result_suffix(char* name) {
     return 0;
 }
 
-// The given result just timed out.
+// A result just timed out.
 // Update the host's avg_turnaround and max_results_day.
 //
 int penalize_host(int hostid, double delay_bound) {