* Keep the most recent builds for each job.

src/Hydra/script/hydra_update_gc_roots.pl

@@ -5,6 +5,7 @@ use File::Path;
 use File::Basename;
 use Hydra::Schema;
 use Hydra::Helper::Nix;
+use POSIX qw(strftime);
 
 my $db = openHydraDB;
 
@@ -16,40 +17,67 @@ my %roots;
 
 sub registerRoot {
     my ($path) = @_;
-    print "$path\n";
+    #print "$path\n";
 
     mkpath($gcRootsDir) if !-e $gcRootsDir;
 
     my $link = "$gcRootsDir/" . basename $path;
         
-    if (!-e $link) {
+    if (!-l $link) {
         symlink($path, $link)
-            or die "cannot creating symlink in $gcRootsDir to $path";
+            or die "cannot create symlink in $gcRootsDir to $path";
     }
 
     $roots{$path} = 1;
 }
 
 
-# Determine which builds to keep automatically.
-my %pathsToKeep;
-
-# TODO
-
-
-# For finished builds, we only keep the output path, not the derivation.
-foreach my $build ($db->resultset('Builds')->search({finished => 1, buildStatus => 0}, {join => 'resultInfo'})) {
-    if ($build->resultInfo->keep || defined $pathsToKeep{$build->outpath}) {
+sub keepBuild {
+    my ($build) = @_;
+    print "keeping build ", $build->id, " (",
+            strftime("%Y-%m-%d %H:%M:%S", localtime($build->timestamp)), ")\n";
     if (isValidPath($build->outpath)) {
         registerRoot $build->outpath;
     } else {
         print STDERR "warning: output ", $build->outpath, " has disappeared\n";
     }
 }
+
+
+# Go over all jobs in all projects.
+
+foreach my $project ($db->resultset('Projects')->all) {
+
+    foreach my $job ($project->builds->search({},
+        {select => [{distinct => 'attrname'}], as => ['attrname']}))
+    {
+        print "*** looking for builds to keep in ", $project->name, ":", $job->attrname, "\n";
+
+        # Keep the N most recent successful builds for each job and
+        # platform.
+        my @recentBuilds = $project->builds->search(
+            { attrname => $job->attrname
+            , finished => 1
+            , buildStatus => 0 # == success
+            },
+            { join => 'resultInfo'
+            , order_by => 'timestamp DESC'
+            , rows => 3 # !!! should make this configurable
+            });
+
+        keepBuild $_ foreach @recentBuilds;
+    }
 }
 
 
+# Keep all builds that have been marked as "keep".
+print "*** looking for kept builds\n";
+my @buildsToKeep = $db->resultset('Builds')->search({finished => 1, keep => 1}, {join => 'resultInfo'});
+keepBuild $_ foreach @buildsToKeep;
+
+
 # For scheduled builds, we register the derivation as a GC root.
+print "*** looking for scheduled builds\n";
 foreach my $build ($db->resultset('Builds')->search({finished => 0}, {join => 'schedulingInfo'})) {
     if (isValidPath($build->drvpath)) {
         registerRoot $build->drvpath;

src/Hydra/sql/hydra.sql

@@ -340,6 +340,7 @@ create trigger cascadeReleaseSetUpdate
 
 create table ReleaseSetJobs (
     project       text not null,
+    -- !!! urgh: "release" is a reserved keyword in sqlite >= 3.6.8!
     release       text not null,
 
     job           text not null,