So my entire infrastructure is on RHEL 6.7 (except MCU, which is 6.5 because MCU is picky). I personally hate typing, mainly because I really stink at it. In order to make my life easier, I've created start and stop scripts for all my servers. This is especially crucial for servers like the Video Manager, which require startup in a specific order. I chose to do this instead of creating services that auto start, just because I'm a control freak who likes to manually control everything. It's probably helpful to understand that my SSC, Community and DB2 servers all reside on a single VM (not recommended for production of course).
#1: Without further ado, here's my Community/SSC/DB2 startup script
This script, like all my scripts, resides in the "/root" directory with 700 permissions. This ensures it can only be run by root.
So the first section, titled "Kill any leftover pids, is because often when one of our servers shuts down dirty, it leaves pid files laying around. If you don't kill these off, your server will startup in a bad state, if at all. Obviously, if you have anything other than a Sametime server installed, you should make sure the other programs don't also leave pids in the "/opt" directory, lest you discover new and interesting ways to break them when you run this script.
#2: Stopping is slightly different
First, you need to type "quit" in the terminal window where Domino is running. Then you can run the following script to stop. Note that you have to pass in username and password to stop the server.
Again, straightforward. Note the stopping order is the inverse of starting.
#3: Advanced script for other servers
The other servers are not quite so interesting, here's the Advanced script (I have a cluster, and the WAS proxy resides on this node, so there's an extra line to start the proxy).
Stopping follows the same template as the Community script.
#4: Starting SolidDB to ensure Vmgr startup success
You can use the above scripts as general guides for any server you wish, with the exception of the Video Manager (Vmgr) and the MCU. The MCU is (or should be) started automatically. The Vmgr is a stand alone server (not federated to the SSC lke all the other children) so you have no node agent to start, but you do need to make sure SolidDB is started first - it's critical to successful Vmgr startup.
#5: Shutting down Vmgr
Shutting down the Vmgr is unique as well; after running the script you'll have to enter SolidDB credentials (default is admin:admin)
JaredWallace 310000KF1P Visits (8638)
One of the things that immediately frustrated me when I went to install my MCU was the slightly obscure fact that RHEL 6 requires quite a few dependencies. We don't make this very obvious in our documentation, and if you do happen to notice, the link we give you brings you to this technote, "Lis
Now the technote is accurate in what it lists. My problem is that it's not set up to just copy and paste. Moreover, why on earth do we require manually installing this stuff? Why can't we release an RPM that does this automatically?
If you just want to install the dependencies, and get on with your life, here's a single command you can copy and paste (remember to be root, and that this is for Red Hat 6.X only!):
yum install dialog glibc-devel glibc-devel.i686 glibc-common zlib-devel.i686 zlib-devel libgcc.i686 libgcc libs
Now instead of being content with that on my environment, what I decided to do was make a dummy package, that would handle the dependencies for me, as well as ensure that httpd was NOT installed (since that prevents the MCU install from proceeding). If you have plans to install a large number of MCUs, this may make sense to replicate. I know practically nothing about package management, but there is a nifty tool called FPM (effing package management) that is super simple to use. (For details on FPM, see http
What you need is the latest MCU from Fix Central (htt
Steps (perform these on a Red Hat 6.X box):
1) Install FPM
yum install ruby-devel gcc
gem install fpm
2) Use FPM to build the dummy MCU package. Note that specifying 32 bit versions of dependencies ("-d") requires a different syntax than yum.
fpm -s dir -t rpm -n wallaceMCU -d dialog -d glibc-devel -d 'gli
Explanation of options:
-d : dependency
-s : source for the rpm is a directory
-t : target is an rpm file
-n : name of the resulting file
The end result is an rpm file named "wallaceMCU.rpm". You can install this on your MCU server by copying it over and running "yum install wallaceMCU.rpm". That action will install all dependencies, make sure Apache (httpd) is not installed, and will extract the MCU code itself into "/So
Jared Wallace, IBM Software Engineer, Sametime Extended Support