[pgpool-general: 1444] Re: removing nodes from pgpool without restarting
ishii at postgresql.org
Thu Feb 28 11:30:22 JST 2013
> Thank you for the explanation. I was thinking of failover as the just the primary node failing;
> it now makes sense to me that it is any node failing.
> I am trying to have autoscaling with no down time, so it would be nice to have
> reload also remove nodes with status 3 that are no longer in pgpool.conf.
Good point. Let me think if this is doable for next pgpool-II version(3.3).
> Thanks again,
You are welcome!
SRA OSS, Inc. Japan
> On Feb 27, 2013, at 7:22 PM, Tatsuo Ishii wrote:
>>> I have set up pgpool 3.2.3 in master/slave mode with streaming replication and it works fine. Now I want to use autoscaling to add and remove nodes as demand fluctuates. Scaling up works with no problems, but when I scale down, I want to remove the extra nodes from pgpool wiithout restarting. What happens now is I call pcp_detach_node(), which changes the status from 2 to 3 for that node, but leaves it in the pool. I also remove the backend from the config file and reload it, but that has no effect unless you restart pgpool.
>>> 1. Is there a way to just remove a node from pgpool?
>> No(without restarting pgpool).
>>> 2. If not, is it a problem to leave a number of nodes with status 3 in the pool? The nodes themselves will have been terminated.
>> Should be no problem. Leaving those unused nodes in pgpool.conf will
>> cause problem when you restart pgpool though (they will trigger
>> immediate failover).
>>> 3. I see that when I detach a standby node, it calls the failover command - why?
>> Pgpool-II handles the failover event in uniform way. In fact the
>> difference between failover and detach is just the former is triggered
>> by detection of errors, while the latter is triggered by pcp command.
>> Tatsuo Ishii
>> SRA OSS, Inc. Japan
>> English: http://www.sraoss.co.jp/index_en.php
>> Japanese: http://www.sraoss.co.jp
More information about the pgpool-general