<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif">Ok,</div><div class="gmail_default" style="font-family:verdana,sans-serif">thank you very much Tatsuo.</div><div class="gmail_default" style="font-family:verdana,sans-serif"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif">Bye.</div><div class="gmail_default" style="font-family:verdana,sans-serif"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif">Meph</div></div><div class="gmail_extra"><br><div class="gmail_quote">On 27 March 2017 at 15:54, Tatsuo Ishii <span dir="ltr"><<a href="mailto:ishii@sraoss.co.jp" target="_blank">ishii@sraoss.co.jp</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">It looks Pgpool-II works as expected. From your pgpool.log, I found 38<br>
SELECTs were sent to DB node 0, while 70 SELECTs were sent to DB node<br>
1. Many of SELECTs sent to DB node 0 is "SELECT COUNT(*) FROM<br>
pg_catalog.pg_roles" (33 out of 38), which will be always sent to DB<br>
node 0 since it is a query against the system catalog. In summary,<br>
most of user SELECTs were sent to DB node 1 according to the<br>
pgpool.log.<br>
<div class="HOEnZb"><div class="h5"><br>
Best regards,<br>
--<br>
Tatsuo Ishii<br>
SRA OSS, Inc. Japan<br>
English: <a href="http://www.sraoss.co.jp/index_en.php" rel="noreferrer" target="_blank">http://www.sraoss.co.jp/index_<wbr>en.php</a><br>
Japanese:<a href="http://www.sraoss.co.jp" rel="noreferrer" target="_blank">http://www.sraoss.co.<wbr>jp</a><br>
<br>
> Hi Tatsuo,<br>
> as attachments of this email you can find logs and configuration file.<br>
><br>
> Very kind regards.<br>
><br>
><br>
> Meph<br>
><br>
> On 26 March 2017 at 04:38, Tatsuo Ishii <<a href="mailto:ishii@sraoss.co.jp">ishii@sraoss.co.jp</a>> wrote:<br>
><br>
>> > Hi there,<br>
>> > at the moment, I am using pgpool-II as load balancer in master-slave mode<br>
>> > with streaming replication with two postgres nodes. Pgpool is accessed<br>
>> by a<br>
>> > java server application that use a JDBC pool in Payara (a Glassfish fork)<br>
>> > application server.<br>
>> ><br>
>> > During startup phase, the application execute about 1800 queries that are<br>
>> > almost all readonly stored functions, but pgpool send queries almost<br>
>> > exclusively toward primary node. This is the situation after some tries:<br>
>> ><br>
>> ><br>
>> ><br>
>> > postgres@postgres=# show pool_nodes;<br>
>> > node_id | hostname | port | status | lb_weight | role<br>
>> |<br>
>> > select_cnt | load_balance_node | replication_delay<br>
>> > ---------+--------------------<wbr>-----+-------+--------+-------<br>
>> > ----+---------+------------+--<wbr>-----------------+------------<wbr>-------<br>
>> > 0 | <a href="http://citrix2.redevogames.com" rel="noreferrer" target="_blank">citrix2.redevogames.com</a> | 15004 | up | <a href="tel:0.250000" value="+390250000">0.250000</a> |<br>
>> primary |<br>
>> > 9520 | true | 0<br>
>> > 1 | 10.0.0.3 | 5997 | up | 0.750000 |<br>
>> standby |<br>
>> > 10 | false | 0<br>
>> ><br>
>> > The question now is: is it a good idea to use a JDBC pool to connect to<br>
>> > pgpgool or is it better without it?<br>
>> ><br>
>> ><br>
>> > It can be other reason for this behavior of balancer?<br>
>><br>
>> There may be something which could explain this (for example<br>
>> replication delay is too much). Can you show me the pgpool log, along<br>
>> with log_per_node_statement = on?<br>
>><br>
>> Best regards,<br>
>> --<br>
>> Tatsuo Ishii<br>
>> SRA OSS, Inc. Japan<br>
>> English: <a href="http://www.sraoss.co.jp/index_en.php" rel="noreferrer" target="_blank">http://www.sraoss.co.jp/index_<wbr>en.php</a><br>
>> Japanese:<a href="http://www.sraoss.co.jp" rel="noreferrer" target="_blank">http://www.sraoss.co.<wbr>jp</a><br>
>><br>
</div></div></blockquote></div><br></div>