Forum Stats

  • 3,782,726 Users
  • 2,254,680 Discussions
  • 7,880,148 Comments

Discussions

EM12cR2 install fail in omsca at upgradeOpss

BrianPa
BrianPa Member Posts: 255
edited Sep 13, 2012 3:45PM in Enterprise Manager
Anyone successfully installed EM12cR2 yet? I had some trouble in the MDS schema configuration, but made it past that, now I'm stuck with a failure in omsca. I'm running Linux x86-64 against an EE 11.2.0.3 repository database, all on one host, doing a 1-system upgrade from EM12cR1+BP1.

Looks like the first error showing up is:
Starting upgrade of opss configuration and security stores.
Command FAILED, Reason: Null Found Searching JpsBase:cn=dmyroot
oracle.security.jps.JpsException: Null Found Searching JpsBase:cn=dmyroot
That's appearing in $OMS_HOME/cfgtoollogs/omsca/logs(date)/emLogs/sysman/log/secure.log.

Error further reported up the stack as:
INFO: oracle.sysman.top.oms:Securing OMS ......
INFO: oracle.sysman.top.oms:EM Key is secured and is backed up at /oracle/oem/Middleware12cR2/oms/sysman/config/emkey.ora
INFO: oracle.sysman.top.oms:java.lang.Exception: Failed during upgradeOpss
INFO: oracle.sysman.top.oms:	at oracle.sysman.omsca.adapter.wls.OMSWLSAdapter.adapterPostDeployAndReposSetup(OMSWLSAdapter.java:2954)
INFO: oracle.sysman.top.oms:	at oracle.sysman.omsca.framework.OMSGenericAdapter.postDeployAndReposSetup(OMSGenericAdapter.java:327)
INFO: oracle.sysman.top.oms:	at oracle.sysman.omsca.framework.OMSCAUpgradeInstall.execute(OMSCAUpgradeInstall.java:407)
INFO: oracle.sysman.top.oms:	at oracle.sysman.omsca.framework.OMSConfigAssistantDriver.main(OMSConfigAssistantDriver.java:240)
INFO: oracle.sysman.top.oms:OMSCA-ERR:Securing of OMS failed
INFO: oracle.sysman.top.oms:
INFO: oracle.sysman.top.oms:OMSCA-ERR:Post "Deploy and Repos Setup" operations failed. Check the trace file:/oracle/oem/Middleware12cR2/oms/cfgtoollogs/omsca/omsca_20120912121637.log
I tend to get better and faster support here than I do from SRs, so trying here first!



Edit to add -- gave up, backing out the upgrade. I'll wait to see how it goes for some of the experts out there.

Edited by: BrianP on Sep 12, 2012 2:53 PM

Best Answer

Answers

  • The first step is to log an SR and upload all the logs files.


    - $ORACLE_HOME/sysman/log/schemamanager zipped
    - $ORACLE_HOME/cfgtoollogs/ zipped
    - central inventory logs

    Tell me the Sr
  • BrianPa
    BrianPa Member Posts: 255
    Thanks... I backed out the change for now since I was trying it out on production (how I love having hardware snapshots!).

    I'll give it a few days/weeks to see how things go and wait for some blog posts about successful upgrades (guess I won't be the first one this time through). Next time I'll clone things over to a test server, try again, and log the SR if the same issue recurs.

    I was hoping this might have been a known issue with an easy resolution, like clearing out SCHEMA_VERSION_REGISTRY to get through the MDS config.
  • sxkumar
    sxkumar Member Posts: 161
    Thanks for your feedback Brian!

    If you still have access to the logs, please share with us via a SR. We would like to figure out what may have gone wrong since we did not see this error in our internal testing.
  • BrianPa
    BrianPa Member Posts: 255
    edited Sep 12, 2012 3:57PM
    Sorry, I should have saved the logs but didn't.

    It looked like some kind of OPSS confusion with the LDAP root somehow. I poked around in v$sql to try to figure out what OMSCA was doing, it was failing after running the following query under SYSMAN_OPSS:
    SELECT t0.JPS_ATTRS_ID, t0.ATTRVAL, t0.ATTRNAME, t0.JPS_DN_ENTRYID FROM JPS_ATTRS t0, JPS_DN t1 WHERE 
    (((t0.JPS_DN_ENTRYID = t1.ENTRYID) AND (t1.PARENTDN = :1 )) AND EXISTS (SELECT :2  FROM JPS_DN t3, CT_10 t2 WHERE 
    (((t2.JPS_DN_ENTRYID = t3.ENTRYID) AND (t2.ATTRVAL = :3 )) AND (t1.ENTRYID = t3.ENTRYID))) )
    It was pulling "cn=dmyroot" for the oracle.security.jps.ldap.root.name property out of the old 12cR1 home's jps-config.xml file, which seemed... wrong. SYSMAN_OPSS.JPS_DN_ENTRYID had a record with PARENTDN = 'cn=root,', but no "cn=dmyroot".

    It may have been my fault due to the way I dealt with the MDS schema configuration failure -- I followed the steps in http://docs.oracle.com/cd/E24628_01/install.121/e24089/appdx_troubleshooting.htm#CEGFEEJJ, but those weren't enough -- I had to drop SYSMAN_MDS cascade and delete from schema_version_registry where comp_name in ('Metadata Services', 'Oracle Platform Security Services'). Once I did that, retrying the MDS schema configuration finally returned success, but I may have just been hiding some other issue. I also did about five flashback restores and retries while trying to work that one out.

    Now that I have some internal folks interested I just might give it another try tomorrow :)

    Edited by: BrianP on Sep 12, 2012 3:57 PM
  • It good to know that you are giving it another try . Can you keep your env intact in case you run into any issue and drop me an email. I can have quick meeting with you and look into the issue as well.

    Are you upgarding from 10.2.0.5 /11.1 to 12.1.0.2 OR 12.1.0.1 to 12.1.0.2 ?
  • BrianPa
    BrianPa Member Posts: 255
    I will leave everything intact and email you if this pops up tomorrow.

    My upgrade was from 12.1.0.1 with BP1, installed from the post-BP1 installer (not an upgrade) originally set up on a brand new 11.2.0.3 repository DB.
  • BrianPa
    BrianPa Member Posts: 255
    edited Sep 13, 2012 3:45PM
    Hi Akanksha,

    I am repeating my installation now, currently at the repository upgrade step. If or when I hit a failure I will email you my contact info and the steps I took to reach this point.

    You may be interested that another individual has reported getting stuck during OMSCA at http://www.gokhanatil.com/2012/09/enterprise-manager-cloud-control-12c-release-2.html (see the comment from Hemant dated Sep. 13 at 16:04).

    Edit to add: Email sent after MDS Schema Configuration failed.
    Edited again to remove the false "workaround" I had previously used. It's not right, don't use it.

    Edited by: BrianP on Sep 13, 2012 9:31 AM

    Edited by: BrianP on Sep 13, 2012 3:45 PM
  • Akanksha Sheoran Kaler-Oracle
    Akanksha Sheoran Kaler-Oracle Member Posts: 439
    Accepted Answer
    Thanks Brain for letting me know. Since you have emailed me, i will followup with you on that thread . Keep your env intact and lets have a web conf with Dev folks to find the cause of the error.
  • BrianPa
    BrianPa Member Posts: 255
    My upgrade has now completed successfully.

    The root cause of my problem (identified with the excellent help of Akanksha's team) was the OpenSSH X11 forwarding timeout for untrusted connections made between the OMS server and the Cygwin X server on my desktop. After 20 minutes, recent versions of OpenSSH do not permit untrusted X11 connections, even with a valid xauth cookie that was good enough to start the installer. The MDS Schema Configuration step, reached near the end of the upgrade, runs a tool that expects it can successfully connect to an X display and fails if it cannot.

    I followed their advice and re-ran my upgrade using VNC for the GUI, and everything completed flawlessly.

    I've used this setup previously to install a new 12c R1+BP1 OEM and several 11.2.0.3 databases, none of which had a problem, so I didn't think there was any chance that it was something in my desktop environment (not even on the OMS server!) causing the upgrade to fail, but it turns out that was exactly it.

    Bravo to the EM12c team, and thanks!
  • sxkumar
    sxkumar Member Posts: 161
    Thanks Brian! Glad we could help.
This discussion has been closed.