[#199] Missing schema changes when user has no permissions on DB

This fixes many glitches with the schema deployment:
- silently ignoring real failure modes such as user not having
correct permissions on the DB
- a couple of broken upgrade steps
- adds --reset so schema can be fully redeployed

Netdisco/lib/App/Netdisco/DB.pm

@@ -11,7 +11,7 @@ __PACKAGE__->load_namespaces(
 );
 
 our # try to hide from kwalitee
-  $VERSION = 39; # schema version used for upgrades, keep as integer
+  $VERSION = 40; # schema version used for upgrades, keep as integer
 
 use Path::Class;
 use File::Basename;

Netdisco/lib/App/Netdisco/DB/schema_versions/App-Netdisco-DB-17-18-PostgreSQL.sql

@@ -1,5 +1,7 @@
 BEGIN;
 
+ALTER TABLE admin DROP CONSTRAINT IF EXISTS admin_pkey;
+
 ALTER TABLE admin ADD PRIMARY KEY (job);
 
 COMMIT;

Netdisco/lib/App/Netdisco/DB/schema_versions/App-Netdisco-DB-38-39-PostgreSQL.sql

@@ -12,6 +12,4 @@
 
 BEGIN;
 
-DELETE n1.* FROM node n1 INNER JOIN (SELECT mac, switch, port from node GROUP BY mac, switch, port HAVING count(*) > 1) n2 ON n1.mac = n2.mac AND n1.switch = n2.switch AND n1.port = n2.port AND n1.vlan = '0';
-
 COMMIT;

Netdisco/lib/App/Netdisco/DB/schema_versions/App-Netdisco-DB-39-40-PostgreSQL.sql

@@ -0,0 +1,17 @@
+-- clean up node table where vlan = 0 and vlan = <another number>
+-- 
+-- DELETE n1.*
+--   FROM node n1 INNER JOIN
+--     (SELECT mac, switch, port from node
+--       GROUP BY mac, switch, port
+--       HAVING count(*) > 1) n2
+--     ON n1.mac = n2.mac
+--       AND n1.switch = n2.switch
+--       AND n1.port = n2.port
+--       AND n1.vlan = '0';
+
+BEGIN;
+
+DELETE FROM node AS n1 USING (SELECT mac, switch, port from node GROUP BY mac, switch, port HAVING count(*) > 1) n2 WHERE n1.mac = n2.mac AND n1.switch = n2.switch AND n1.port = n2.port AND n1.vlan = '0';
+
+COMMIT;

Netdisco/lib/App/Netdisco/Manual/Deployment.pod

@@ -177,6 +177,13 @@ following commands into a shell script and call it nightly from C<cron>:
 This will keep 30 days of backups. You don't need to stop Netdisco during the
 backup.
 
+=head1 Database Schema Redeployment
+
+The database schema can be fully redeployed (even over an existing
+installation, in a safe way) using the following command:
+
+ ~netdisco/bin/netdisco-db-deploy --reset
+
 =head1 Further Reading...
 
 Other ways to run and host the web application can be found in the

Netdisco/lib/App/Netdisco/Manual/ReleaseNotes.pod

@@ -36,7 +36,25 @@ but they are backwards compatible.
 
 =back
 
-=head1 2.031000
+=head1 2.031003
+
+=head2 Health Advice
+
+This release will I<once again> remove from the database spurious Node
+(workstation, printer, etc) entries on vlan 0, which were causing dupliate
+entries in the web interface. We advise that you back up the database prior to
+upgrade:
+
+ /usr/bin/pg_dump -F p --create -f netdisco-pgsql.dump netdisco
+
+=head2 General Notices
+
+=head1 2.031003
+
+The database schema can be fully redeployed (even over an existing
+installation, in a safe way) using the following command:
+
+ ~netdisco/bin/netdisco-db-deploy --reset
 
 =head2 General Notices