[Pgcluster-general] S.O.S., replication sever failover (solved)

["Tomás A. Rossi"]

Leif Madsen escribió:
> On 1/8/07, "Tomás A. Rossi" <tomas at mecon.gov.ar> wrote:
>   
>>  Thanks, it all works now. I've discovered that it all was a problem a one
>> of the nodes missing some entry in pg_hba.conf.
>>  The weird thing is that, for the failover to work properly, I had to add
>> the IP of the node in the pg_hba.conf at that node.
>>
>>  I mean:
>>
>>  The IPs are:
>>  ldapserver: 10.11.22.22
>>  ldapserver2: 10.11.80.80
>>
>>  pg_hba.conf at ldapserver2 was:
>>  ...
>>  # IPv4 local connections:
>>  host    all         all         127.0.0.1/32            trust # <-- not
>> enough for local pgreplicate to connect.
>>  host    all         all         10.11.22.22/32          trust # ldapserver
>>  ...
>>
>>  And I had to add the entry:
>>  host    all         all         10.11.80.80/32          trust # ldapserver2
>> (and now local pgreplicate can connect to local postmaster).
>>
>>  ...which I thought wasn't necessary since 127.0.0.1/32 was present in the
>> ACL. Who knows :)
>>     
>
> I think you're looking for the following entry for local:
>
> local    all    all                                            trust
> # ldapserver
Nope, in fact, that entry was there already from the very beginning. 
That's for Unix domain socket connections.
Thanks anyway,
--
Tom;
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgcluster-general/attachments/20070109/58156c2b/attachment.html