Skip to Content

CMS down during upgrade

I am performing an upgrade on a quite complex system:

it’s composed by 8 hosts:

  1. 2 hosts for the Application Tier (Tomcats)
  2. 2 hosts for the Intelligence Tier (CMS, FRS, connection servers, event server)
  3. 2 hosts for Processing Tier (AJS, APS, Crystal, Dashboard, Webi server)
  4. 2 hosts for Lumira

CMS database is on another host.

There are no CMS running on hosts 1,3 and 4.

Starting version is BO 4.1 SP7 patch 3 + Design Studio 1.6 + Lumira 1.3

Final version will be BO 4.2 SP4 patch 2 hotfix 1 + Lumira 2.0 SP2

Upgrade paths are:

  • BO 4.1 SP7 patch 3 -> BO 4.2 SP4
  • BO 4.2 SP4 -> patch 2
  • Hotfix 1
  • Lumira 1.30 -> Lumira 2.0 SP2

I upgraded a first platform (a “sandbox” made as similar as possibile to the Production system).

Everything was fine on hosts 1 and 2 (application and intelligence tier).

When I run the upgrades on hosts reported with 3 and 4 in the previous list (specifically the hosts on which there are BO processing server but the CMS) it happened a strange thing: the phase “waiting for cms to start” hanged (every single time) because CMS (both) switched to stopped state !

Obviously, at the beginning of the upgrade they were up&running otherwise upgrade didn't start. Fortunately was sufficient to start CMS manually and the upgrade resumed and completed.

Why this behavior ?

There is a particular path to upgrade such a complex system ?

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

3 Answers

  • Oct 03, 2017 at 11:45 AM

    1. You only need one CMS server running in complex cluster during upgrade.
    2. The steps are basically, stop everything on every host, except one CMS, upgrade that CMS machine. If good - install upgrade on the rest of the hosts at the same time, without starting anything on them.
    3. The problem with both cms's stopping while upgrade is being run on another host is in-usual and requires investigation.
    Maybe it wasn't related to upgrade ?
    In any case - the CMS logs from both server and install/upgrade log from the machine being upgraded should be reviewed.

    Add comment
    10|10000 characters needed characters exceeded

  • avatar image
    Former Member
    Oct 03, 2017 at 01:13 PM

    Hi Denis

    thank you very much for you answer.

    I'm glad to know that the procedure I followed was the right one !

    I am performing other tests in order to clarify if it happens again !

    Thank you !

    Andrea

    Add comment
    10|10000 characters needed characters exceeded

  • Oct 10, 2017 at 08:13 PM

    Hi everybody

    finally I set up a laboratory on virtual machine to reproduce steps... and I get the issue again !

    I set up a simple laboratory composed only by two nodes: "winboxi" and "secondo".

    Both on Windows 2012 R2.

    BO starting version is 4.1 SP7 patch 3; only on "secondo" is installed Lumira 1.30 as well.

    winboxi runs these processing servers:

    on host "winboxi" are installed Tomcat and cms database (SQL Anywhere) as well.

    secondo these:

    (CentralManagementServer is present but stopped, disabled and NOT autostarted).

    The steps I performed were:

    1. shutdown "secondo";
    2. updated "winboxi" from BO 4.1 SP7 patch 3 to BO 4.2 SP4 (everything goes fine);
    3. accessed to CMC, removed old 4.1 license and added new 4.2 license;
    4. started update (the same as point 2) on node "secondo";
    5. GOT THE ISSUE !

    At a certain point I noticed that update hangs in "waiting for cms to start" phase; so I tried to access CMC but I get an error at logon screen. Tried to retrieve the status of processing servers from ccm console on the server (SIA for "winboxi" proved to be in "running state") but I got an error after credential input.

    So I restarted SIA on "winboxi" manually... and, after a copule of minutes, the update on "secondo" went on until the end ("successfully with warnings" but everything seems to be ok).

    I attached two log files:

    • SIA_WINBOXI_trace.000001.glf.txt
    • cms_WINBOXI.CentralManagementServer_trace.000001.glf.txt

    copied in the moment I got the issue; hope that someone finds something intersting in them !

    Please help me !

    Add comment
    10|10000 characters needed characters exceeded

    • Hi Denis

      thank you very much for you comment.

      I'm starting to think that the problem is REALLY on that stopped and not-autostarting CMS.

      In this case, is there a way to delete only the CMS of a node ? Or Maybe would be better starting it before the upgrades ?