[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: I'd consider this bug very serious



found 514132 95
found 472627 94lenny1
thanks

Hello again,

Martin Pitt [2009-02-05 23:57 +0100]:
> TBH this is not really the time to fiddle with it, for Lenny I'd just
> back out the original change completely.

Done and uploaded, debdiff attached.

@release team: This is a direct revert of the change that went into
version 94, so it's veeery safe.

> The check was introduced to help users who misconfigured their
> clusters to run on the same port, and thus got hangs on shutdown,
> etc. But that's not anywhere near a grave problem.

I reopened the original bug (#472627) for Lenny, and keep this bug
open for the experimental version, so that I won't forget about it.

Bernd's patch looks fine to me, but before I apply it to the
experimental branch, I want to create a test case for this regression,
and change the existing test case for #472627 to match the new patch.
I'm not going to do that past midnight, and don't want to do it "under
the gun" for Lenny. Let's stay on the safe side this close to the
release.

Thanks to everyone involved, and good night,

Martin
-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)
diff -Nru postgresql-common-94/debian/changelog postgresql-common-94lenny1/debian/changelog
--- postgresql-common-94/debian/changelog	2008-12-06 20:20:00.000000000 +0100
+++ postgresql-common-94lenny1/debian/changelog	2009-02-06 00:06:23.000000000 +0100
@@ -1,3 +1,13 @@
+postgresql-common (94lenny1) unstable; urgency=high
+
+  * Urgency high because of RC bug fix (regression from Etch).
+  * pg_ctlcluster, t/030_errors.t: Back out colliding ports check
+    (r896), since in its current form it breaks setups where multiple
+    clusters listen to the same port number, but are using different
+    socket directories. This reopens #472627. (Closes: #514132)
+
+ -- Martin Pitt <mpitt@debian.org>  Fri, 06 Feb 2009 00:06:00 +0100
+
 postgresql-common (94) unstable; urgency=low
 
   * t/070_non_postgres_clusters.t: Test that all cluster configuration files
diff -Nru postgresql-common-94/pg_ctlcluster postgresql-common-94lenny1/pg_ctlcluster
--- postgresql-common-94/pg_ctlcluster	2008-12-01 11:46:45.000000000 +0100
+++ postgresql-common-94lenny1/pg_ctlcluster	2009-02-06 00:02:22.000000000 +0100
@@ -243,14 +243,6 @@
 
     start_check_pid_file;
 
-    # check conflicting port
-    foreach my $v (get_versions) {
-        foreach my $c (get_version_clusters $v) {
-            error("Port conflict: cluster $v/$c is already running on port " .
-                $info{'port'}) if cluster_port_running $v, $c, $info{'port'};
-        }
-    }
-
     # get locale used by initdb
     my ($lc_ctype, $lc_collate) = get_cluster_locales $version, $cluster;
     $lc_ctype or error ('Could not parse locale out of pg_controldata output');
diff -Nru postgresql-common-94/t/030_errors.t postgresql-common-94lenny1/t/030_errors.t
--- postgresql-common-94/t/030_errors.t	2008-12-01 11:46:45.000000000 +0100
+++ postgresql-common-94lenny1/t/030_errors.t	2009-02-06 00:02:22.000000000 +0100
@@ -6,7 +6,7 @@
 
 use lib 't';
 use TestLib;
-use Test::More tests => 162;
+use Test::More tests => 154;
 
 use lib '/usr/share/postgresql-common';
 use PgCommon;
@@ -237,22 +237,6 @@
     'pg_ctlcluster start succeeds again with reappeared /var/lib/postgresql';
 is_program_out 'postgres', "pg_ctlcluster $version main stop", 0, '', 'stopping cluster';
 
-# pg_ctlcluster checks colliding ports
-ok ((system "pg_createcluster $version other >/dev/null") == 0,
-    "pg_createcluster other");
-set_cluster_port $version, 'other', '5432';
-is ((exec_as 'postgres', "pg_ctlcluster $version main start"), 0,
-    'pg_ctlcluster: main cluster on conflicting port starts');
-like_program_out 'postgres', "pg_ctlcluster $version other start", 1,
-    qr/conflict.*8.3\/main/, 
-    'pg_ctlcluster other cluster on conflicting port fails';
-is_program_out 'postgres', "pg_ctlcluster $version main stop", 0, '', 
-    'stopping main cluster';
-is ((exec_as 'postgres', "pg_ctlcluster $version other start"), 0,
-    'pg_ctlcluster: other cluster on conflicting port starts after main is down');
-ok ((system "pg_dropcluster $version other --stop") == 0, 
-    'pg_dropcluster other');
-
 # clean up
 ok ((system "pg_dropcluster $version main") == 0, 
     'pg_dropcluster');

Attachment: signature.asc
Description: Digital signature


Reply to: