[pgpool-general: 1994] Re: Problem with watchdog...

[Jeff Frost]

Correcting my cc'ing of the wrong mailing list.

On Aug 2, 2013, at 1:39 PM, Jeff Frost <jeff at pgexperts.com> wrote:

> So are you indicating that your test is:
> 
> killall -9 pgpool 
> 
> on the primary pgpool server?
> 
> I think I may have lost track of the sequence of events.  Maybe you could lay it out as a simple test case for us?
> 
> On Aug 2, 2013, at 1:24 PM, Fernando Buzon <fbuzon at creddefense.com> wrote:
> 
>> log on the active: (nothing changes after kill command)
>> 
>> 2013-08-02 17:01:17 LOG:   pid 11860: wd_create_send_socket: connect() reports failure (Connection timed out). You can safely ignore this while starting up.
>> 2013-08-02 17:01:29 LOG:   pid 11860: wd_escalation: escalated to master pgpool
>> 2013-08-02 17:02:32 LOG:   pid 11860: wd_create_send_socket: connect() reports failure (Connection timed out). You can safely ignore this while starting up.
>> 2013-08-02 17:02:32 LOG:   pid 11860: wd_escalation:  escalated to delegate_IP holder
>> 2013-08-02 17:02:32 LOG:   pid 11860: wd_init: start watchdog
>> 2013-08-02 17:02:32 ERROR: pid 11948: read_startup_packet: incorrect packet length (0)
>> 2013-08-02 17:02:32 ERROR: pid 11949: read_startup_packet: incorrect packet length (0)
>> 2013-08-02 17:02:32 LOG:   pid 11860: pgpool-II successfully started. version 3.2.5 (namameboshi)
>> 2013-08-02 17:02:32 LOG:   pid 11860: find_primary_node: primary node id is 1
>> 
>> log on the standby: (nothing changes too)
>> 
>> 2013-08-02 17:22:05 LOG:   pid 6844: wd_chk_sticky: ifup[/sbin/ifconfig] doesn't have sticky bit
>> 2013-08-02 17:22:08 LOG:   pid 6844: wd_create_send_socket: connect() reports failure (No route to host). You can safely ignore this while starting up.
>> 2013-08-02 17:22:08 LOG:   pid 6844: wd_init: start watchdog
>> 2013-08-02 17:22:08 LOG:   pid 6844: pgpool-II successfully started. version 3.2.5 (namameboshi)
>> 2013-08-02 17:22:08 LOG:   pid 6844: find_primary_node: primary node id is 1
>> 
>> 
>> OBS
>> only works well if I get the job done correctly with "pgpool stop". Then...
>> 
>> 2013-08-02 17:23:52 LOG:   pid 6846: wd_escalation: escalated to master pgpool
>> 2013-08-02 17:23:55 LOG:   pid 6846: wd_create_send_socket: connect() reports failure (No route to host). You can safely ignore this while starting up.
>> 2013-08-02 17:23:55 LOG:   pid 6846: wd_escalation:  escalated to delegate_IP holder
>> 
>> 
>> 2013/8/2 Fernando Buzon <fbuzon at creddefense.com>
>> I am using pgpool 3.2.5
>> 
>> The logs do not change anything.
>> I interrupt all active server processes at once with the command: "killall -9 pgpool"
>> If I restart the standby pgpool or former active pgpool, everything works normally again.
>> 
>>  
>> 
>> 
>> 2013/8/2 Jeff Frost <jeff at pgexperts.com>
>> 
>> On Aug 2, 2013, at 1:05 PM, Fernando Buzon <fbuzon at creddefense.com> wrote:
>> 
>> > I configured wd_hostname with IP from the eth0 like you said but still the same problem.
>> >
>> > someone has already done this test?
>> > stop the active pgpool with killall -9 pgpool?
>> > what would happen in this case?
>> > who would bring down interface eth0:0 (delegate_ip) if pgpool is no longer running on the server?
>> >
>> 
>> 
>> Can you post your logs before and after the stop?
>> 
>> Also, what pgpool processes are still running?
>> 
>> Yes, it's expected to work.
>> 
>> You might have mentioned before, but what version of pgpool are you running?
>> 
>> 
>> 
> 
> ---
> Jeff Frost <jeff at pgexperts.com>
> CTO, PostgreSQL Experts, Inc.
> Phone: 1-888-PG-EXPRT x506
> FAX: 415-762-5122
> http://www.pgexperts.com/ 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Pgp mailing list
> Pgp at lists.pgexperts.com
> http://lists.pgexperts.com/mailman/listinfo/pgp

---
Jeff Frost <jeff at pgexperts.com>
CTO, PostgreSQL Experts, Inc.
Phone: 1-888-PG-EXPRT x506
FAX: 415-762-5122
http://www.pgexperts.com/ 






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.pgpool.net/pipermail/pgpool-general/attachments/20130802/4fa684b7/attachment.htm>