Allow collect of bind query generating temporary files. Until now only

temporary file statistics was reported but not the query generating
these file. This collect is possible only if log_connection and
log_disconnection are activated in postgresql.conf. Thanks to Ildefonso
Camargo for the feature request.

diff --git a/pgbadger b/pgbadger
index 37f8eb943d427d2ada635acd48cee45eef46b7fa..c4961bccfd2c7f97656b0f9d19e9063d644d5c6a 100755 (executable)
--- a/pgbadger
+++ b/pgbadger
@@ -354,6 +354,7 @@ my @log_files   = ();
 my %prefix_vars = ();
 my $q_prefix    = '';
 my @prefix_q_params = ();
+my %last_execute_stmt = ();
 
 my $remote_host = '';
 my $ssh_command = '';
@@ -13392,6 +13393,7 @@ sub validate_log_line
                # Log line does not match the required dbname
                if (!$prefix_vars{'t_dbname'} || !grep(/^$prefix_vars{'t_dbname'}$/i, @dbname)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13400,6 +13402,7 @@ sub validate_log_line
                # Log line does not match the required dbuser
                if (!$prefix_vars{'t_dbuser'} || !grep(/^$prefix_vars{'t_dbuser'}$/i, @dbuser)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13408,6 +13411,7 @@ sub validate_log_line
                # Log line does not match the required dbclient
                if (!$prefix_vars{'t_client'} || !grep(/^$prefix_vars{'t_client'}$/i, @dbclient)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13416,6 +13420,7 @@ sub validate_log_line
                # Log line does not match the required dbname
                if (!$prefix_vars{'t_appname'} || !grep(/^\Q$prefix_vars{'t_appname'}\E$/i, @dbappname)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13424,6 +13429,7 @@ sub validate_log_line
                # Log line matches the excluded db
                if ($prefix_vars{'t_dbname'} && grep(/^\Q$prefix_vars{'t_dbname'}\E$/i, @exclude_db)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13432,6 +13438,7 @@ sub validate_log_line
                # Log line matches the excluded dbuser
                if ($prefix_vars{'t_dbuser'} && grep(/^$prefix_vars{'t_dbuser'}$/i, @exclude_user)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13440,6 +13447,7 @@ sub validate_log_line
                # Log line matches the excluded appname
                if ($prefix_vars{'t_appname'} && grep(/^\Q$prefix_vars{'t_appname'}\E$/i, @exclude_appname)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13449,6 +13457,7 @@ sub validate_log_line
                # Log line matches the excluded dbclient
                if ($prefix_vars{'t_client'} && grep(/^$prefix_vars{'t_client'}$/i, @exclude_dbclient)) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        return 0;
                }
        }
@@ -13512,6 +13521,11 @@ sub parse_query
        my $cur_last_log_timestamp = "$prefix_vars{'t_year'}-$prefix_vars{'t_month'}-$prefix_vars{'t_day'} " .
                                     "$prefix_vars{t_hour}:$prefix_vars{t_min}:$prefix_vars{t_sec}";
 
+       # Delete old bind query temporary storage
+       foreach my $p (keys %last_execute_stmt) {
+               delete $last_execute_stmt{$p} if ($p > $t_pid);
+       }
+
        # Force some LOG messages to be ERROR messages so that they will appear
        # in the event/error/warning messages report.
        if ($prefix_vars{'t_loglevel'} eq 'LOG') {
@@ -13552,12 +13566,14 @@ sub parse_query
        # Remove session failure from current workload because there is no disconnection entry
        if (!$disable_session && ($prefix_vars{'t_loglevel'} eq 'FATAL')) {
                delete $current_sessions{$prefix_vars{'t_pid'}};
+               delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                if ($is_tsung_output) {
                        delete $tsung_session{$prefix_vars{'t_pid'}}
                }
        } elsif (!$disable_session && ($prefix_vars{'t_loglevel'} eq 'WARNING')) {
                if ($prefix_vars{'t_query'} =~ /terminating connection/) {
                        delete $current_sessions{$prefix_vars{'t_pid'}};
+                       delete $last_execute_stmt{$prefix_vars{'t_pid'}};
                        if ($is_tsung_output) {
                                delete $tsung_session{$prefix_vars{'t_pid'}}
                        }
@@ -13700,7 +13716,9 @@ sub parse_query
                $overall_stat{$curdb}{'peak'}{$cur_last_log_timestamp}{tempfile_size} += $size;
                $overall_stat{$curdb}{'peak'}{$cur_last_log_timestamp}{tempfile_count}++;
                if ($fmt eq 'csv') {
-                       $cur_temp_info{$t_pid}{query}     = $prefix_vars{'t_statement'};
+                       $cur_temp_info{$t_pid}{query} = $prefix_vars{'t_statement'};
+               } elsif (exists $last_execute_stmt{$t_pid}) {
+                       $cur_temp_info{$t_pid}{query} = $last_execute_stmt{$t_pid}{query};
                }
                # Stores information related to first created temporary file
                if (!exists $cur_temp_info{$t_pid}{timestamp}) {
@@ -13790,6 +13808,7 @@ sub parse_query
                $host = _gethostbyaddr($host) if ($dns_resolv);
 
                delete $current_sessions{$prefix_vars{'t_pid'}};
+               delete $last_execute_stmt{$prefix_vars{'t_pid'}};
 
                if ($is_tsung_output) {
                        $tsung_session{$prefix_vars{'t_pid'}}{disconnection}{date} = $prefix_vars{'t_timestamp'};
@@ -14121,6 +14140,13 @@ sub parse_query
                $overall_stat{$curdb}{histogram}{query_time}{$k}++;
                $overall_stat{$curdb}{histogram}{query_total}++;
                $prefix_vars{'t_bind'} = 1;
+               # Stores execute query in case of temporary file
+               # only if log_disconnection is enabled otherwise
+               # the risk is to fill memory with this storage.
+               if (scalar keys %session_info) {
+                       $last_execute_stmt{$prefix_vars{'t_pid'}}{query} = $prefix_vars{'t_query'};
+                       $last_execute_stmt{$prefix_vars{'t_pid'}}{timestamp} = $cur_last_log_timestamp;
+               }
        # Activate storage of the explain plan generated by auto_explain
        } elsif ($prefix_vars{'t_query'} =~ s/duration: ([0-9\.]+) ms\s+plan://s) {
                $cur_plan_info{$prefix_vars{'t_pid'}}{duration} = $1;
@@ -14144,6 +14170,8 @@ sub parse_query
                # Skipping parse and bind logs
                return if ($t_action !~ /query|statement|execute/);
                $prefix_vars{'t_bind'} = 1;
+               # Stores execute query in case of temporary file
+               $last_execute_stmt{$prefix_vars{'t_pid'}} = $prefix_vars{'t_query'};
        # Log line that could not be parsed
        } elsif ($is_log_level) {
                if ($prefix_vars{'t_query'} !~