Messaging Server 3.6x NscpMail process continues to grow until memory used up

Question

After upgrading to Messaging Server version 3.6, I have found that the
NscpMail process
continues to grow until I run out of memory and the system eventually crashes.
<P>
One solution to this problem is to remove the problematic messages.<BR>
<P>
<OL>
<LI>Shut down the Messaging Server (see details
below).
<P>
<LI>Move all of the message directories to a temporary location
(see details below).
<P>
<LI>Create these directories again for Messaging Server (see details below).
<P>
<LI>Re-start the Messaging Server (see details below)
<P>
<LI>Examine the message files in the old queue (from Step 1) for
corrupt headers.
<P>
<LI>After you have checked the messages, manually forward the message bodies
to the intended recipients (see details below).
</OL>
<P>
<HR>
<P>
How messages are stored:<BR>
<P>
Prior to delivery, mail messages are separated into three files:
control, body
, and header
.<BR>
<P>
The body and
header files are stored in
the following directory:<BR>
<P>
<I>server root</I>/messages
<P>
The control files are
stored in the following directories:<BR>
<P>
<I>server root</I>/control<BR>
<I>server root</I>/deferred/Error-Handler<BR>
<I>server root</I>/deferred/Account-Handler<BR>
<I>server root</I>/deferred/SMTP-Router<BR>
<P>
Removing messages from the queue and
re-introducing them later:</A><BR>
<P>
<UL>
<LI>Shut down the server using ./NscpMail stop
<BR>
<P>
%cd /etc<BR>
%./NscpMail stop<BR>
<P>
<LI>Move the directories to a temporary location.</A><BR>
<P>
%cd <I>server root</I>/postoffice
%mv control control.sav<BR>
%mv messages message.sav<BR>
%mv deferred deferred.sav<BR>
%mv hold hold.sav<BR>
<P>
<LI>Create these directories again for the Messaging Server.
</A><BR>
<P>
%mkdir control messages deferred hold<BR>
%chown <I>mail-user</I> control messages deferred
hold<BR>
%chgrp <I>mail-group</I> control messages deferred hold
<BR>
<P>
<LI>Start the Messaging Server using ./NscpMail
start.</A><BR>
<P>
%cd /etc<BR>
%./NscpMail start<BR>
<P>
<LI>Forward the message bodies:</A><BR>
<P>
After you have checked all of the messages for corrupt headers, you can copy
the messages back into the queue, at most 1000 at a time.<BR>
<P>
<UL>
<LI>Use the cp command
to copy files instead of the mv
command.
<P>
<LI>Copy the body and
header files into their
respective directories first. Then, copy the control
files into the control
directory. Copying the
control files back into the
control directory will
cause the corresponding messages to be processed the next time the server
flushes the queue, which is done every hour.
<P>
<LI>When you copy the control
files, make sure that you also copy
the corresponding body
and header files.
Otherwise, you will receive errors that say the body
and header
files cannot be opened.
<P>
<LI>To copy groups of body
and header
files and their corresponding
control files, you can
use the cp command with
wildcards. For instance, to copy all the messages that arrived between 12:30
and 12:40 back into the queue, you can use the following commands:<BR>
<P>
%cd <I>server root</I>/postoffice/messages.sav<BR>
%cp 199080325123* <I>server root</I>
/postoffice/messages<BR>
%cd <I>server root</I>/postoffice/control.sav<BR>
%cp 199080325123* <I>server root</I>
/postoffice/control<BR>
%cd <I>server root</I>/postoffice/deferred.sav<BR>
%cp 199080325123* <I>server root</I>
/postoffice/control<BR>
%cd <I>server root</I>/postoffice/hold.sav<BR>
%cp 199080325123* <I>server root</I>/postoffice/hold<BR>
</UL>

Answer 1

Next time when this happen it will be great if you can create statspack snapshots during the problem. Then you will have more insight information of what's going on in Oracle.