MultiXactID Wrap-Around - Mailing list pgsql-hackers

ops, i did forget to update the e-mail subject, sorry. I am reposting it with an appropriate one.<br /><br
/><b><i></i></b>--------------------------------------------------------------------------------<blockquote
class="replbq"style="border-left: 2px solid rgb(16, 16, 255); margin-left: 5px; padding-left: 5px;"><span
style="font-family:times new roman;">I'm keeping on studying multixact.c and log management, and I hope you can help
me,as usual, in clearing my doubts.</span><br style="font-family: times new roman;" /><br style="font-family: times new
roman;"/><span style="font-family: times new roman;">My doubts now concern MultixactID wrap-around management.
</span><brstyle="font-family: times new roman;" /><span style="font-family: times new roman;">Afaics, it is possible to
spawnmultixactids so quickly to have a wrap-around and to start overwriting the data stored in the offset slru (but
analogousconsiderations apply to the member slru as well). This would cause corruption, if the overwritten info was
stillneeded, e.g., by a (very) long-running transaction. This is of course very unlikely in practice, but yet still
possiblein theory.</span><br style="font-family: times new roman;" /><br style="font-family: times new roman;" /><span
style="font-family:times new roman;">In GetNewMultiXactId () wrap-around of MultiXactId seems to be simply handled this
way:</span><brstyle="font-family: times new roman;" /><pre class="fragment" style="font-family: times new roman;">00780
   <span class="comment">_/* Handle wraparound of the nextMXact counter */</span><br />00781     <span
class="keywordflow">if</span>(<a class="code"
href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8c.html#a16"target="_blank">MultiXactState</a>-><a
class="code"href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/structMultiXactStateData.html#o0"
target="_blank">nextMXact</a>< <a class="code"
href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8h.html#a1"target="_blank">FirstMultiXactId</a>)<br
/>00782        <a class="code" href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8c.html#a16"
target="_blank">MultiXactState</a>-><aclass="code"
href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/structMultiXactStateData.html#o0"target="_blank">nextMXact</a>
=<a class="code" href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8h.html#a1"
target="_blank">FirstMultiXactId</a>;</pre><brstyle="font-family: times new roman;" /><span style="font-family: times
newroman;">I cannot see how this may avoid possible overwriting of still needed <span style="text-decoration:
underline;">data.</span>To address such an issue shouldn't one need to check against </span><a class="code"
href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8c.html#a17"style="font-family: times new roman;"
target="_blank">OldestMemberMXactId</a><spanstyle="font-family: times new roman;">, </span><a class="code"
href="http://www.mcknight.de/pgsql-doxygen/cvshead/html/multixact_8c.html#a18"style="font-family: times new roman;"
target="_blank">OldestVisibleMXactId?</a>Or, alternatively, rely on an approach similar to the one taken to handle
standardXID generation (xidWarnLimit, see GetNewTransactionId)?<span style="font-family: monospace;"><span
style="text-decoration:underline;"><br /><br /></span></span>Is it me who's missing something or is it just that such a
casehas been considered so unlikely not to motivate additional overheads/checks?<br /><br />Thanks in advance!<br /><br
/>    Paolo</blockquote><br /><p> Chiacchiera con i tuoi amici in tempo reale! <br />
http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com