From f26d6b664022701ed7a4a6b6c391493b1e54434f Mon Sep 17 00:00:00 2001
From: hjp <hjp@d47eeb7a-11fe-0310-b519-e5a4e12f9765>
Date: Sat, 9 Nov 2019 10:22:34 +0000
Subject: [PATCH] Run backups for different filesets in parallel

---
 MYMETA.json     |  6 ++--
 MYMETA.yml      |  2 +-
 lib/Simba/CA.pm | 85 ++++++++++++++++++++++++++++++++++++-------------
 scripts/backup  |  7 +++-
 4 files changed, 72 insertions(+), 28 deletions(-)

diff --git a/MYMETA.json b/MYMETA.json
index b795006..d0e7d4e 100644
--- a/MYMETA.json
+++ b/MYMETA.json
@@ -4,13 +4,13 @@
       "unknown"
    ],
    "dynamic_config" : 0,
-   "generated_by" : "Module::Build version 0.422",
+   "generated_by" : "Module::Build version 0.4224",
    "license" : [
       "perl_5"
    ],
    "meta-spec" : {
       "url" : "http://search.cpan.org/perldoc?CPAN::Meta::Spec",
-      "version" : "2"
+      "version" : 2
    },
    "name" : "Simba",
    "prereqs" : {
@@ -50,5 +50,5 @@
       ]
    },
    "version" : "0.002",
-   "x_serialization_backend" : "JSON::PP version 2.27300_01"
+   "x_serialization_backend" : "JSON::PP version 2.97001"
 }
diff --git a/MYMETA.yml b/MYMETA.yml
index 9aed85b..d507fb5 100644
--- a/MYMETA.yml
+++ b/MYMETA.yml
@@ -4,7 +4,7 @@ author:
   - unknown
 build_requires: {}
 dynamic_config: 0
-generated_by: 'Module::Build version 0.422, CPAN::Meta::Converter version 2.150005'
+generated_by: 'Module::Build version 0.4224, CPAN::Meta::Converter version 2.150010'
 license: perl
 meta-spec:
   url: http://module-build.sourceforge.net/META-spec-v1.4.html
diff --git a/lib/Simba/CA.pm b/lib/Simba/CA.pm
index 07c8ea5..4a99828 100644
--- a/lib/Simba/CA.pm
+++ b/lib/Simba/CA.pm
@@ -87,33 +87,26 @@ sub new {
     $self->{log_level} = 99;
     $self->{record_file_id} = 0;
     $self->{record_time} = 0;
-    if ($opt->{dbi}) {
-        $self->{dbh} = DBI->connect(@{ $opt->{dbi} },
-                                       { AutoCommit => 0,
-                                         PrintError => 1, 
-                                         RaiseError => 1
-                                       }
-                           );
-    } elsif ($opt->{dbi_file}) {
+    $self->{parallel} = $opt->{parallel};
+
+    if ($opt->{dbi_file}) {
         my $fn = $opt->{dbi_file};
         open(FN, "<$fn") or die "cannot open $fn: $!";
         my $line = <FN>;
         close(FN);
         my @cred = split(/[\s\n]+/, $line);
-        $self->{dbh} = DBI->connect(@cred,
-                                       { AutoCommit => 0,
-                                         PrintError => 1, 
-                                         RaiseError => 1
-                                       }
-                           );
-        $self->{instances_part_size} = 10_000_000;
-        $self->adjust_partitions;
-    } elsif ($opt->{tokyocabinet}) {
-        my $tdb = $self->{tdb} = TokyoCabinet::TDB->new();
-        $tdb->open($opt->{tokyocabinet}, $tdb->WRITER, $tdb->OCREAT)
-            or die "open $opt->{tokyocabinet} failed: " . $tdb->errmsg($tdb->ecode());
+        $self->{dbi} = \@cred;
     }
-    # XXX - DBI
+
+    $self->{dbh} = DBI->connect(@{ $self->{dbi} },
+                                   { AutoCommit => 0,
+                                     PrintError => 1, 
+                                     RaiseError => 1
+                                   }
+                       );
+    $self->{instances_part_size} = 10_000_000;
+    $self->adjust_partitions;
+
     $self->{targets} = $self->{dbh}->selectall_arrayref("select * from filesets", { Slice => {} });
     if ($opt->{filesets}) {
         $self->{targets} =
@@ -144,13 +137,59 @@ sub new {
     return $self;
 }
 
+
 sub run {
     my ($self) = @_;
 
     # run sequentially for prototype. In production we probably
     # want some concurrency
-    for my $target (@{$self->{targets}}) {
-	$self->backup2disk($target);
+    if ($self->{parallel}) {
+        $self->{dbh}->disconnect();
+        my %running = (); 
+        for my $target (@{$self->{targets}}) {
+            $self->log(3, "found target host " . $target->{host} . " dir " . $target->{dir});
+            while (scalar keys %running >= $self->{parallel}) {
+                $self->log(3, "reached parallel limit - waiting");
+                my $pid = wait();
+                delete $running{$pid};
+                $self->log(3, "child with pid $pid terminated, " . (scalar keys %running) . " remaining");
+            }
+            my $pid = fork();
+            if (!defined($pid)) {
+                die "fork failed: $!";
+            }
+            if ($pid == 0) {
+                $self->{dbh} = DBI->connect(@{ $self->{dbi} },
+                                               { AutoCommit => 0,
+                                                 PrintError => 1, 
+                                                 RaiseError => 1
+                                               }
+                                   );
+                $self->backup2disk($target);
+                $self->{dbh}->disconnect();
+                exit(0);
+            } else {
+                $running{$pid} = 1;
+                $self->log(3, "child with pid $pid started, " . (scalar keys %running) . " running");
+            }
+            sleep(10);
+        }
+        while (scalar keys %running) {
+            my $pid = wait();
+            delete $running{$pid};
+            $self->log(3, "child with pid $pid terminated, " . (scalar keys %running) . " remaining");
+        }
+        $self->{dbh} = DBI->connect(@{ $self->{dbi} },
+                                       { AutoCommit => 0,
+                                         PrintError => 1, 
+                                         RaiseError => 1
+                                       }
+                           );
+
+    } else {
+        for my $target (@{$self->{targets}}) {
+            $self->backup2disk($target);
+        }
     }
     $self->log(3, "statistics:");
     for (sort keys %{ $self->{counts} }) {
diff --git a/scripts/backup b/scripts/backup
index 8fcca32..76d788a 100755
--- a/scripts/backup
+++ b/scripts/backup
@@ -9,7 +9,11 @@ use File::stat;
 
 my @filesets;
 
-GetOptions('filesets=i' => \@filesets);
+my $parallel;
+GetOptions(
+    'filesets=i' => \@filesets,
+    'parallel=i' => \$parallel
+);
 @filesets = split(/,/,join(',',@filesets));
 
 $ENV{PATH} = "/usr/bin";
@@ -22,6 +26,7 @@ my $ca = Simba::CA->new({
 			    dbi_file => $ENV{SIMBA_DB_CONN} || "$ENV{HOME}/.dbi/simba",
 			    fh_log   => $log,
 			    (@filesets ? ( filesets => \@filesets ) : ()),
+                            parallel => $parallel,
 			});
 $ca->log_level(9);