<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hi All,<br>
Last weeks with increased number of cores and limited number of
connections to caldb we had problems with database access during
simulation and cooking.<br>
Marty Wise (dbase manager) set up the process to kill stale idle
connections in the
event scheduler. To run the clas software accessing the caldb
(gsim,a1,user_ana) we have to recompile it with the <br>
new caldb/Map available from SVN (fixed by Nathan, see below),
otherwise you may run into problem with reconnection crashes.<br>
Harut<br>
<br>
<br>
<br>
On 10/2/14, 9:53 AM, Nathan Baltzell wrote:<br>
</div>
<blockquote cite="mid:C9586BC8-607A-48E4-9FA4-F4739ABFCEE4@jlab.org"
type="cite">
<pre wrap="">Hi Harut,
The io/caldb/Map in SVN is ready.
I reverted my latest change to go back to one persistent connection instead of
hammering the server with many connect+disconnects. But the bugfix on the
interpretation of mysql_ping return value will allow us to automatically reconnect
when server kills connections that have been sleeping for a long time.
-Nathan
</pre>
</blockquote>
------- Original Message --------
<table class="moz-email-headers-table" cellpadding="0"
cellspacing="0" border="0">
<tbody>
<tr>
<th align="RIGHT" nowrap="nowrap" valign="BASELINE">Subject: </th>
<td>[Clas_offline] caldb timeout/reconnect bug</td>
</tr>
<tr>
<th align="RIGHT" nowrap="nowrap" valign="BASELINE">Date: </th>
<td>Thu, 11 Sep 2014 09:59:09 -0400</td>
</tr>
<tr>
<th align="RIGHT" nowrap="nowrap" valign="BASELINE">From: </th>
<td>Nathan Baltzell <a class="moz-txt-link-rfc2396E" href="mailto:baltzell@jlab.org"><baltzell@jlab.org></a></td>
</tr>
<tr>
<th align="RIGHT" nowrap="nowrap" valign="BASELINE">To: </th>
<td>clas offline <a class="moz-txt-link-rfc2396E" href="mailto:clas_offline@jlab.org"><clas_offline@jlab.org></a></td>
</tr>
</tbody>
</table>
<br>
<br>
<pre>FYI Everyone,
After changes in JLab firewalls on August 16, at least a couple run groups (eg6 and eg4)
started getting new segfaults related to timeout errors in connections to the calibration
database during cooking. While CLAS already had a mechanism in place to automatically
reconnect in such cases, it had a bug resulting in false success that led to the recent
problems.
Correctly testing the return value from mysql_ping fixes it, and it’s now in svn’s
trunk/io/caldb/Map/calib_connect.c (line 122), as well as cvs’s
packages/caldb/Map/calib_connect.c (line 113) just in case.
This bug only (?) became relevant after a new 30 minute timeout on dormant connections
implemented in the JLab firewall on 8/16. Maybe there will be related issues.
Regards,
Nathan
_______________________________________________
Clas_offline mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Clas_offline@jlab.org">Clas_offline@jlab.org</a>
<a class="moz-txt-link-freetext" href="https://mailman.jlab.org/mailman/listinfo/clas_offline">https://mailman.jlab.org/mailman/listinfo/clas_offline</a>
</pre>
</body>
</html>