[pmg-devel] [PATCH pmg-api 1/2] backup: reorder tables in order of dependency

[Stoiko Ivanov]

Currently we dump the tables manually and print the relevant
DDL statements into the dump-file before copying the data.

The complete backup and dump specific tables part would probably be
best replaced by an appropriate `pg_dump` invocation - but as a
stop-gap measure reorder the tables in order of their dependencies.

Noticed while adding a foreign key constraint on 2 new tables
(objectgroup_attribute and rule_attribute) and running backup+restore.

Reported-by: Dominik Csapak <d.csapak at proxmox.com>
Signed-off-by: Stoiko Ivanov <s.ivanov at proxmox.com>
---
 src/PMG/Backup.pm | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/PMG/Backup.pm b/src/PMG/Backup.pm
index 5891afb..ab7e628 100644
--- a/src/PMG/Backup.pm
+++ b/src/PMG/Backup.pm
@@ -90,18 +90,18 @@ sub dumpdb {
     eval {
 	$dbh->begin_work;
 
 	# read a consistent snapshot
 	$dbh->do("SET TRANSACTION ISOLATION LEVEL SERIALIZABLE");
 
+	dump_table($dbh, 'objectgroup', $ofh, 'objectgroup_id_seq', 'id');
+	dump_table($dbh, 'object', $ofh, 'object_id_seq', 'id');
 	dump_table($dbh, 'attribut', $ofh);
 	dump_table($dbh, 'objectgroup_attributes', $ofh);
-	dump_table($dbh, 'rule_attributes', $ofh);
-	dump_table($dbh, 'object', $ofh, 'object_id_seq', 'id');
-	dump_table($dbh, 'objectgroup', $ofh, 'objectgroup_id_seq', 'id');
 	dump_table($dbh, 'rule', $ofh, 'rule_id_seq', 'id');
+	dump_table($dbh, 'rule_attributes', $ofh);
 	dump_table($dbh, 'rulegroup', $ofh);
 	dump_table($dbh, 'userprefs', $ofh);
 
 	# we do not save the following tables: cgreylist, cmailstore, cmsreceivers, clusterinfo
     };
     my $err = $@;
-- 
2.39.2