<p>Well</p>
<p> </p>
<p>I've did a test preventing autvacuum vacuuming Slony's tables using both master and slave</p>
<p> ALTER TABLE _mycluster.sl_nodelock SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_setsync SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_table SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_sequence SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_node SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_listen SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_path SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_subscribe SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_set SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_event SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_confirm SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_seqlog SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_registry SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_seqlastvalue SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_config_lock SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_archive_counter SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_status SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_log_1 SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_log_2 SET (autovacuum_enabled=off);</p>
<p>and with vac_frequency = 3 (both on master and slave in slon configuration file)</p>
<p>As a consequence Slony, is doing the "vacuum analyze" himself</p>
<p>Here is what I see on the log (slave)</p>
<p>2011-09-21 13:18:24 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_archive_counter;<br />2011-09-21 13:18:24 CEST[11540] INFO cleanupThread: 0.648 seconds for vacuuming<br />2011-09-21 13:50:59 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_event;<br />2011-09-21 13:50:59 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_confirm;<br />2011-09-21 13:50:59 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_setsync;<br />2011-09-21 13:50:59 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_seqlog;<br />2011-09-21 13:51:00 CEST[11540] DEBUG1 cleanupThread: vacuum analyze "_mycluster".sl_archive_counter;<br />2011-09-21 13:51:00 CEST[11540] INFO cleanupThread: 0.697 seconds for vacuuming</p>
<p>So for vac_frequency = 3, this is the response to my question :)</p>
<p>I will wait the end of the test to see if I've got a deadlock or not :)</p>
<p><br /><br />On Wed, 21 Sep 2011 10:31:48 +0200, david.techer wrote:<br />
<blockquote style="padding-left:5px; border-left:#1010ff 2px solid; margin-left:5px; width:100%">
<p>Hi</p>
<p>I've begun to test Slony 2.0.7 a couple days ago.</p>
<p> </p>
<p>I've read the Slony documentation and see on the source for clean_thread.c for vac_frequency</p>
<p>However, I've got a question</p>
<p>What happens if I put both on master and slave</p>
<p>1)</p>
<p> ALTER TABLE _mycluster.sl_event SET (autovacuum_enabled=off);<br /> ALTER TABLE _mycluster.sl_confirm SET (autovacuum_enabled=off);</p>
<p> </p>
<p>2) vac_frequency=3 or vac_frequency=0 ?</p>
<p>For your information, I use PostgreSQL 8.4, Slony 2.0.7. Autovacuum is on both master and slave.</p>
<p> </p>
<p>If I have understood,</p>
<p>vac_frequency=0 implies that SLony will not used his "own" maintenance operation and just do a analyze if required</p>
<p> </p>
<p>Let me know.</p>
<p><span id="_rc_sig"></span></p>
<pre>--------------------------------------
Jean David TECHER
davidgis.fr
--------------------------------------</pre>
<p> </p>
<pre>--------------------------------------
Jean David TECHER
davidgis.fr
--------------------------------------</pre>
<p> </p>
</blockquote>
</p>