Hi,
I recall, too, that I had to temporarily switch to trust which I
disliked doing. The attached email might help you out if ident
authentication is an option for you.
-ar
On 09/26/2012 10:02 AM, Raphael Nadeau wrote:
Hello
I just installed bucardo on my test environment and I am hitting a
wall. The problem I have is this error :
MCP Warning: Killed (line 44): DBI
connect('dbname=xxxxx;port=5432;host=xxxxxxxxxxxx','bucardo',...)
failed: fe_sendauth: no password supplied at
/usr/share/perl5/Bucardo.pm line 265
(I have masked the dbname and host, but this is basically it)
I am pretty sure I configured Postgresql right. I am listening to all
interface and I have made sure that bucardo is an accepted user from
the specified IP address, but I did not set it to "trust", I have
setted it to md5.
Also, I have made pgpass files that the postgres process can read (or
should read)
The whole installation was done on Debian 6.0.5 with apt-get, so I
have bucardo 4.4.0-1
I do not understand what is happening. Normally, the password
provided in the pgpass does not seem to be taken into consideration.
Thank you for your help
Raphaël
Raphaël Nadeau, ing. / Eng.
Chef Concepteur Électrique / Chief Electrical Designer
Eocycle Technologies inc.
49 du Bel-Air #106
Levis, Qc, Canada
G6V 6K9
T. (418) 833-0926 x.222
F. (418) 833-8152
_______________________________________________
Bucardo-general mailing list
[email protected]
https://mail.endcrypt.com/mailman/listinfo/bucardo-general
--- Begin Message ---
Hi,
our source database is in the same cluster as the bucardo database.
Hence, I received an error when issuing
bucardo_ctl add table <blah> db=<yadda>
because the validate_goat function executes as whoever's running the
database, thus potentially conflicting with the user as specified in the
connection string as returned by bucardo.db_getconn.
What do you think of the following patch?
--- bucardo.schema.org 2012-06-11 16:35:23.000000000 -0500
+++ bucardo.schema 2012-06-13 19:19:59.000000000 -0500
@@ -850,6 +850,8 @@
$row = $rv->{rows}[0]{apple};
($db{dsn},$db{user},$db{pass},$db{ssp}) = split /\n/ => $row;
+$db{user} = '' unless $db{dsn} =~ /host=/;
+
elog(DEBUG, "Connecting in validate_goat as $db{dsn} user=$db{user} pid=$$
for table $schema.$table");
$dbh = DBI->connect($db{dsn}, $db{user}, $db{pass},
It essentially removes the username in those cases is which the host name
is missing in the connection string which in turn implies utilization of
UNIX sockets. Thoughts?
Cheers,
-ar
--- End Message ---
_______________________________________________
Bucardo-general mailing list
[email protected]
https://mail.endcrypt.com/mailman/listinfo/bucardo-general
