From f44c9910e704f3e932fdcdbd7a1f9359d36e2be3 Mon Sep 17 00:00:00 2001
From: David Anderson <davea@ssl.berkeley.edu>
Date: Tue, 12 Jul 2011 20:44:28 +0000
Subject: [PATCH] - validator: if job FLOPs estimates are accurate,     PFC
 values should be around 1.     If they differ from 1 by a factor of > 1e4,
 ignore them,     and put an error message into the validator log - validator:
 if get_pfc() fails because an app version is     missing from the DB (i.e.
 the project deleted it)     keep going so we don't reprocess the WU forever

svn path=/trunk/boinc/; revision=23837
---
 checkin_notes    | 12 ++++++++++++
 sched/credit.cpp | 33 +++++++++++++++++++++++++--------
 2 files changed, 37 insertions(+), 8 deletions(-)

diff --git a/checkin_notes b/checkin_notes
index 1542094805..af0fd6a523 100644
--- a/checkin_notes
+++ b/checkin_notes
@@ -4066,3 +4066,15 @@ David  12 July 2011
             app_details.php
         inc/
             boinc_db.inc
+
+David  12 July 2011
+    - validator: if job FLOPs estimates are accurate,
+        PFC values should be around 1.
+        If they differ from 1 by a factor of > 1e4, ignore them,
+        and put an error message into the validator log
+    - validator: if get_pfc() fails because an app version is
+        missing from the DB (i.e. the project deleted it)
+        keep going so we don't reprocess the WU forever
+
+    sched/
+        credit.cpp
diff --git a/sched/credit.cpp b/sched/credit.cpp
index 16d2810b31..48171871a6 100644
--- a/sched/credit.cpp
+++ b/sched/credit.cpp
@@ -337,6 +337,17 @@ inline double wu_estimated_credit(WORKUNIT& wu, DB_APP& app) {
     return wu_estimated_pfc(wu, app)*COBBLESTONE_SCALE;
 }
 
+inline bool is_pfc_sane(double x, WORKUNIT& wu, DB_APP& app) {
+    if (x > 1e4 || x < 1e-4) {
+        log_messages.printf(MSG_CRITICAL,
+            "Bad FLOP ratio (%f): check workunit.rsc_fpops_est for %s (app %s)\n",
+            x, wu.name, app.name
+        );
+        return false;
+    }
+    return true;
+}
+
 // Compute or estimate "claimed peak FLOP count".
 // Possibly update host_app_version records and write to DB.
 // Possibly update app_version records in memory and let caller write to DB,
@@ -638,7 +649,10 @@ int get_pfc(
                 raw_pfc, wu.rsc_fpops_est
             );
         }
-        avp->pfc_samples.push_back(raw_pfc/wu.rsc_fpops_est);
+        double x = raw_pfc / wu.rsc_fpops_est;
+        if (is_pfc_sane(x, wu, app)) {
+            avp->pfc_samples.push_back(x);
+        }
     }
 
     if (config.debug_credit) {
@@ -649,11 +663,11 @@ int get_pfc(
             (r.received_time - r.sent_time)
         );
     }
-                
-    hav.pfc.update(
-        raw_pfc / wu.rsc_fpops_est,
-        HAV_AVG_THRESH, HAV_AVG_WEIGHT, HAV_AVG_LIMIT
-    );
+
+    double x = raw_pfc / wu.rsc_fpops_est;
+    if (is_pfc_sane(x, wu, app)) {
+        hav.pfc.update(x, HAV_AVG_THRESH, HAV_AVG_WEIGHT, HAV_AVG_LIMIT);
+    }
     hav.et.update_var(
         r.elapsed_time / wu.rsc_fpops_est,
         HAV_AVG_THRESH, HAV_AVG_WEIGHT, HAV_AVG_LIMIT
@@ -727,7 +741,7 @@ int assign_credit_set(
             log_messages.printf(MSG_CRITICAL,
                 "get_pfc() error: %s\n", boincerror(retval)
             );
-            return retval;
+            continue;
         } else {
             if (config.debug_credit) {
                 log_messages.printf(MSG_NORMAL,
@@ -766,9 +780,12 @@ int assign_credit_set(
     double x;
     if (normal.size()) {
         x = low_average(normal);
-    } else {
+    } else if (approx.size()) {
         x = vec_min(approx);
+    } else {
+        x = 0;
     }
+
     x *= COBBLESTONE_SCALE;
     if (config.debug_credit) {
         log_messages.printf(MSG_NORMAL,