<p>No one knows what DB_BUFFER_SMALL mean on Fedora DS?</p>
<p>How does a change were not found in the changelog? A problem? A bug? Something "usual"?</p>
<p>Is there anything that could prevent this error?</p>
<p>"DANIEL CRISTIAN CRUZ" <daniel.cruz@sc.senai.br> escreveu:<br />
 </p>
<blockquote type="cite">
<p>Hi all,</p>
<p>I had some error messages that I don't understood:</p>
<pre>
[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - changelog program - agmt="cn=m1-m2-xxx" (m2:389): Failed to retrieve change with CSN 4857f6db000000010000; db error - -30999 <br />DB_BUFFER_SMALL: User memory too small for return value<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - agmt="cn=m1-m2-xxx" (m2:389): A changelog database error was encountered<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - changelog program - agmt="cn=m1-c1-xxx" (c1:389): Failed to retrieve change with CSN 4857f6db000000010000; db error - -30999 DB_BUFFER<br />_SMALL: User memory too small for return value<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - agmt="cn=m1-c1-xxx" (c1:389): A changelog database error was encountered<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - changelog program - agmt="cn=m1-c2-xxx" (c2:389): Failed to retrieve change with CSN 4857f6db000000010000; db error - -30999<br /> DB_BUFFER_SMALL: User memory too small for return value<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - agmt="cn=m1-c2-xxx" (c2:389): A changelog database error was encountered<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - changelog program - agmt="cn=m1-c3-xxx" (c3:389): Failed to retrieve change with CSN 4857f6db000000010000; db error - -30999 DB_<br />BUFFER_SMALL: User memory too small for return value<br />[17/Jun/2008:14:39:11 -0300] NSMMReplicationPlugin - agmt="cn=m1-c3-xxx" (c3:389): A changelog database error was encountered</pre>
<p>My schema has two multi masters: m1 and m2, and many consumers c1, c2, c3, ..., cX.</p>
<p>There are replication agreements between m1 and m2.</p>
<p>There are replication agreements from each master to all consumers.</p>
<p>It's the second time I saw the problem, and I need to make a full initialize from m1 to all others, where the log above were taken from m1. m1 receive updates from m2, even after this error.</p>
<p>After initializing m2 from m1: I saw the message on m2:</p>
<pre>
[17/Jun/2008:15:02:04 -0300] NSMMReplicationPlugin - replica_reload_ruv: Warning: new data for replica ou=foo,o=bar does not match the data in the changelog.<br /> Recreating the changelog file. This could affect replication with replica's  consumers in which case the consumers should be reinitialized.</pre>
<p>In a test environment I didn't got this problem.</p>
<p>Is there something wrong in my structure (two masters and many consumers)?</p>
<p>Is any configuration missing (some gotchas)?</p>
<p>Any help would be apreciated.</p>
<p>Regards,</p>
</blockquote><hr style="border:dotted 1px silver; width:90%; border:dotted 1px silver;"><p><span style="color: #000080"><strong>Daniel Cristian Cruz</strong><br />
</span><strong>Administrador de Banco de Dados<br />
<strong>Direção Regional - </strong>Núcleo de Tecnologia da Informação<br />
SENAI - SC<br />
Telefone: 48-3239-1422 (ramal 1422)</strong></p>