OES In Place Upgrade Thinks Server Is New

I know there is a long thread about OES upgrade issues and honestly, I thought this one was resolved with the latest/blind release of the updated OES 23.4 ISO.

I've been upgrading a bunch of OES 2018.3 servers to OES 24.2.  These are all physical servers, bare metal installs.

I did about four of them with absolutely no issue, just booting the latest and greatest 23.4 ISO off a USB key.  On the fifth one, I ran into a problem.

During the eDirectory upgrade portion after the first reboot, the options looked different.  I didn't have a sever IP/hostname selection box.  I figured this was because the server has replicas on it. 

I tried to proceed with the eDir upgrade, but it would just fail.  I discovered looking at the details it was trying to do an "ndsconfig add", as if this was a brand new server.

I SSH'd into the box and could see that eDir was fine. I manually ran an ndsconfig upgrade and it worked fine.  eDirectory is healthy and functioning.  

At some point, I gave up on the eDirectory upgrade screen and the server rebooted and is up.  But OES is not right.  If I run a yast2 oes-install it still thinks the server is NOT in the tree, it wants to do an ndsconfig add.

Is there a trick to get the OES install to recognize that this server was an upgrade, not a fresh install?

On top of that, I discovered that I had some NSS volumes on an iSCSI connection that was not connected during the install.  The NSS Pool is 32 bit.  What is the manual process to upgrade that?

Thanks.

Matt

  • 0  
    Is there a trick to get the OES install to recognize that this server was an upgrade, not a fresh install?

    Actually remembering to select Upgrade at the beginning iso boot, rather than Install option.  I did that once as well, but caught it early enough (those eDir questions) to roll back and start again.

    But you are past that, so this might just be time to open a case, as you are clearly in tread carefully territory.

    Assuming that you can see the server still in the tree, and that the other boxes have the same iSCSI type connections. (bits support will need to know)

    ________________________

    Andy of KonecnyConsulting.ca in Toronto
    Please use the "Like" and/or "Verified Answers" as appropriate as that helps us all.

  • 0   in reply to   

    Most definitely picked upgrade. It was the normal upgrade screens and selections up to this point.  

    I do have a case open, but so far it's been silence other than a request for a support config. Support isn't exactly stellar these days.

    eDir is fine, as I mentioned, I manually ran the ndsconfig upgrade process and verified eDir health.

    But if I try and run "yast2 oes-install" now it wants to do an "ndsconfig add" like it doesn't even think the server is in the the tree, which it clearly is.

    Matt

  • Suggested Answer

    0  
    But if I try and run "yast2 oes-install" now it wants to do an "ndsconfig add" like it doesn't even think the server is in the the tree, which it clearly is.

    hmm,  does eDir think it was configured?  check all the applicable parts of OES for the =YES or =NO with

    grep CONFIGURED= /etc/sysconfig/novell/*

    <grrrrr> We have seen many instances where those aren't updated properly. </grrrrr>

    how is the basic eDir health check going (-T,   -E,    -C -A -Ad) ?

    ________________________

    Andy of KonecnyConsulting.ca in Toronto
    Please use the "Like" and/or "Verified Answers" as appropriate as that helps us all.

  • Suggested Answer

    0   in reply to   

    So that was the key, most of the services looked like this:

    edir:SERVICE_CONFIGURED="no"
    edir_oes2018:SERVICE_CONFIGURED="yes"
    edir_oes2018_sp1:SERVICE_CONFIGURED="yes"
    edir_oes2018_sp2:SERVICE_CONFIGURED="yes"
    edir_oes2018_sp3:SERVICE_CONFIGURED="no"

    I went in and manually changed all the services there were previously configured to yes.

    Now when I ran "yast2 oes-install" I got the already configured services listing.

    From here I reconfigured the services (except for eDir) and it configured everything and rebooted the server and now it seems fine.

    I also figured out my NSS pool issue. It had nothing to do with it being 32 bit.  I didn't build this server originally but I noticed that there were two paths for the iSCSI connection.  Apparently they are REDUNDANT paths and I was connecting BOTH of them.  Once I disabled one and left one connected I was able to mount the pool and the volumes no problem.

    So I think I'm good now, I believe the server is functioning properly.

    Thanks for the tip.

    Matt