Problems with EM Database Control after applying 10.2.0.4 patchset

Question

Hello all,
Host: Linux x86
DB: Single-instance running EM Database Control. Originally 10.2.0.1 patched to 10.2.0.4
Last night I upgraded the database from 10.2.0.1 to 10.2.0.4. All appeared to go well. I didn't get any errors in the process and when I logged in to the database control, all appeared well. It was late at night so I didn't browse around for very long, though. When I came in this morning and did some more extensive testing, upon logging in I first noticed that in the General section the database version was listed as 10.2.0.1.0. I thought that this was pretty odd since I had just been logged on to SQL*Plus and saw that the database was reporting back that it was at version 10.2.0.4.0. I refreshed the page and I saw
"java.lang.Exception: IOException in reading Response :: Connection reset"
at the top of the page. I have seen this problem before and, in the past, it has been resolved by either restarting the dbconsole or making sure that the monitoring configuration is set up correctly. The first thing that I did was issue
[host]$ emctl stop dbconsole followed by
[host]$ emctl start dbconsole
to see if that would do the trick. When I saw that it didn't, I went into the Monitoring Configuration and was surprised to see that it said "No items were found." I refreshed the page and this error appeared:
oracle.sysman.emSDK.emd.comm.CommException: IOException in reading Response :: Connection reset
I tried to follow all of the instructions in the ReadMe as closely as I could. I have been searching on the Internet and the Oracle Forums for most of the day to try to find out the cause of this error. I see that people have these errors, but the circumstances are not the same as mine. Is this a common problem? Does anyone know what I can do to resolve it?
Any help would be greatly appreciated.
Thanks!
John

Answer 1

Hello Satish,
Thank you for your reply. I don't know if this is normal or not but, now after the upgrade, in the ORACLE_HOME I have a directory that's been created called hostnameSID (the fully qualified host name of my computer and the SID of the database).  Inside of it is a sysman directory, an emctl.pid file and an oraInst.loc file.  I also still have a syman directory under the ORACLEHOME but the dbconsole appears to be running out of the hostname_SID directory.  Is that normal?
I looked up the article that you referenced and looked for the targets.xml file first in the ORACLE_HOME/sysman/emd directory. I found a targets.xml file along with several other similar files, targets.xml.1, targets.xml.2, etc. The targets.xml file just had the following:
<Targets>
</Targets>
Seeing that it was blank, I then took a look at the the hostname_SID/sysman/emd folder. I saw only one targets.xml file in this location and when I opened it up it had the following:
<Targets AGENT_TOKEN="+alpha numeric string+">
<Target TYPE="oracle_emd" NAME="+hostname+:3938"/>
<Target TYPE="host" NAME="+hostname+"/>
</Targets>
The contents of that file seemed to match what the article which you cited indicated should be in the file. Since I have a single instance managed by the dbconsole and not EM Grind Control, I'm expecting the files to be a little bit different. Am I wrong in assuming that? Still, I'm not sure if I should have the hostname_SID directory beneath the ORACLE_HOME.
Any ideas?
Thanks again for your help,
John