On Fri, Aug 19, 2011 at 8:55 PM, Craig James <craig_james@emolecules.com> wrote:
> On 8/19/11 1:40 AM, Marko Kreen wrote:
>>
>> On Fri, Aug 19, 2011 at 2:44 AM, Craig James<craig_james@emolecules.com>
>> wrote:
>>>
>>> We had to temporarily assign two IP addresses to our servers. After
>>> doing
>>> so and rebooting, Londiste will start, but it just sits there doing
>>> nothing.
>>> The logfile has zero bytes, and it doesn't seem to connect to either the
>>> master or the slave database.
>>>
>>> We reconfigured Postgres to listen on both addresses, but the /etc/hosts
>>> tables still point to the original addresses (that is, the second IP
>>> address
>>> shouldn't matter). But I can connect to Postgres on either IP address on
>>> all servers.
>>>
>>> All other Postgres applications work. I can use psql, and our Perl DBI
>>> and
>>> PHP applications all work with no problems.
>>>
>>> When I invoke "londiste.py --verbose ...", the log file says:
>>>
>>> 2011-08018 16:35:47,250 22074 DEBUG Attaching
>>>
>>> And that's all. Nothing more ever happens.
>>>
>>> Help? Thanks!
>>
>> Does the exact connect string you use in londiste config work
>> with psql? Like so:
>>
>> $ psql -d 'connstr'
>
> As I mentioned earlier, yes it does work.
>
> I also discovered by looking at all the postgres processes that it's making
> the first connection (to the master), but it doesn't seem to even try to
> make the second connection.
>
> Could there be something else it's waiting for? The dual-network issue may
> be completely irrelevant. I know that while the IT guy was reconfiguring
> the network, he had to do a forced-reboot at least once without shutting
> down Postgres. Is it possible that the Londiste tables are in some invalid
> state, and the daemon is waiting for something that's never going to happen?
Look at the logs, what queries does it do?
--
marko