Changeset 31490 for gs3-extensions

Timestamp:

2017-03-10T22:03:47+13:00 (7 years ago)

Author:

ak19

Message:

1. Fix to issue of a tomcat host/port change not propagating to solr host/port change when rebuilding a solr collection after tomcat host/port change. The change to tomcat server props need to be made after gs3-setup.sh was already run in the terminal earlier, to encouner the problem upon solr build. The bug was reproduced on Linux, and the fix for it also tested on Linux. Still need to test fix out on Windows. 2. Simultaneously made http protocol used in solr more robust to whether it's http or https.

Location:

gs3-extensions/solr/trunk/src

Files:

• bin/script/solr_passes.pl (modified) (1 diff)
• perllib/solrserver.pm (modified) (1 diff)
• perllib/solrutil.pm (modified) (2 diffs)

gs3-extensions/solr/trunk/src/bin/script/solr_passes.pl

@@ -67 +67 @@
 
   # Now start up the solr-post command
-  &solrutil::open_post_pipe($core);
+  &solrutil::open_post_pipe($core, $solr_server->get_solr_base_url());
 }
 

gs3-extensions/solr/trunk/src/perllib/solrserver.pm

@@ -43 +43 @@
     $self->{'server_explicitly_started'} = undef;
 
-    my $server_port = $ENV{'SOLR_PORT'};
-    my $server_host = $ENV{'SOLR_HOST'};
-    my $base_url = "http://$server_host:$server_port/solr/";
-    my $admin_url = "http://$server_host:$server_port/solr/admin/cores";
-    
-    $self->{'base-url'} = $base_url;
-    $self->{'admin-url'} = $admin_url;
+    # set SOLR_HOST and SOLR_PORT env vars (tomcat host and port, if not using jetty) 
+    # by calling ant get-default-solr-servlet if possible. Else fallback on whatever the existing env vars are.
+    # tomcat host and port would have been set up in the env as SOLR_HOST and SOLR_PORT
+    # In case someone changed the tomcat host/port, we want to update the solr server variables too
+    my $solr_url = &solrutil::get_solr_servlet_url();    
+    # get the url parts, though we won't be using most of them
+    my ($protocol, $server_host, $server_port, $servlet_name) = &solrutil::get_solr_url_parts($solr_url);
+    
+    # set the solr server env vars to what was discovered, so that any other old perl code
+    # dependent on these env vars will have any changes propagated. 
+    # (All perl code referencing these env vars should already be updated, but still...)
+    $ENV{'SOLR_HOST'} = $server_host;
+    $ENV{'SOLR_PORT'} = $server_port;
+    
+    $self->{'base-url'} = $solr_url; # e.g. of the form http://localhost:8383/solr
+    $self->{'admin-url'} = "$solr_url/admin/cores";
 
     return bless $self, $class;
+}
+
+sub get_solr_base_url {
+    my $self = shift (@_);
+    return $self->{'base-url'};
 }
 

gs3-extensions/solr/trunk/src/perllib/solrutil.pm

@@ -59 +59 @@
 }
 
+# The get-solr-servlet-url ant target can be run from anywhere by specifying the
+# location of GS3's ant build.xml buildfile.
+# GSDL3SRCHOME will be set for GS3 by gs3-setup.sh.
+# Based on servercontrol::get_library_URL.
+sub get_solr_servlet_url {
+    # Set up fall backs, incl. old way of using solr host and port values that's already in the environment
+    my $solr_url = "http://".$ENV{'SOLR_HOST'}.$ENV{'SOLR_PORT'}."/solr"; # fallback to default
+
+    my $perl_command = "ant -buildfile \"$ENV{'GSDL3SRCHOME'}/build.xml\" get-solr-servlet-url";
+    
+    if (open(PIN, "$perl_command |")) {
+    while (defined (my $perl_output_line = <PIN>)) {
+        if($perl_output_line =~ m@(https?):\/\/(\S*)@) { # grab all the non-whitespace chars
+        $solr_url="$1://".$2; # preserve the http protocol
+        }
+    }
+    close(PIN);
+    
+    #print STDERR "XXXXXXXXXX SOLR URL: $solr_url\n";
+
+    } else {
+    print STDERR "*** ERROR IN solrutil::get_solr_servlet_url:\n";
+    print STDERR "    Failed to run $perl_command to work out GS3's solr URL\n";
+    print STDERR "    falling back to using original solr_URL: $solr_url\n";
+    }
+
+    return $solr_url;
+}
+
+# Given the solr base url (e.g. http://localhost:8383/solr by default), this function
+# returns the url's parts: protocol, host, port, solr servlet
+sub get_solr_url_parts {
+    my $solr_url = shift (@_);
+
+    # Set up fall backs, incl. old way of using solr host and port values that's already in the environment
+    my ($protocol, $server_host, $server_port, $servlet_name)
+    = ("http://", $ENV{'SOLR_HOST'}, $ENV{'SOLR_PORT'}, "solr");
+
+    
+    # http://stackoverflow.com/questions/8206135/storing-regex-result-in-a-new-variable
+    if($solr_url =~ m@(https?)://([^:]*):([0-9]*)/(.*)$@) { # m@https?://([^:]*):([^/])/(.*)@) {
+    
+    ($protocol, $server_host, $server_port, $servlet_name) = ($1, $2, $3, $4);
+    
+    #print STDERR "XXXXXXXXXX PROTOCOL: $protocol, SOLR_HOST: $server_host, SOLR_PORT: $server_port, servlet: $servlet_name\n";
+
+    } else {
+    print STDERR "*** WARNING: in solrutil::get_solr_url_parts(): solr servlet URL not in expected format\n";
+    }
+
+    return ($protocol, $server_host, $server_port, $servlet_name);
+}
 
 
 sub open_post_pipe
 {
-    my ($core) = @_;
+    my ($core, $solr_base_url) = @_;
 
     my $search_path = get_search_path();
@@ -72 +124 @@
     my $full_post_jar   = solrutil::locate_file($search_path,$post_jar);
     
-    my $server_port = $ENV{'SOLR_PORT'}; # tomcat
-    my $server_host = $ENV{'SOLR_HOST'};
-    
     # Now run solr-post command
-    my $post_props = "-Durl=http://$server_host:$server_port/solr/$core/update";
+    my $post_props = "-Durl=$solr_base_url/$core/update"; # robustness of protocol is taken care of too
+
     $post_props .= " -Ddata=stdin";
     $post_props .= " -Dcommit=yes";