Remotely starting the CMS server using a SSH script in a Unix environment.

[Former Member]

Is it possible to remotely start a CMS server using a shell script which starts an SSH session and runs the startservers command asin the command below:

ssh -o ConnectionAttempts=2 -o ConnectTimeout=10 -l boadmin huxd0226 '/opt/BO/boxir2/bobje/startservers'

The command securely logs into the server (huxd0226) and runs the command described in single quotes.

the CMS server actually appears to start but then dies inexplicably and displays no errors in the trace logs. The full log file is displayed below:-

Timestamp ProcessID ThreadID Message

Thu Sep 18 10:59:05 2008 913548 1 (wireobinit.cpp:97): trace messa

ge: Creating the shared CWireObject::PropertyNameMap

Thu Sep 18 10:59:05 2008 913548 1 (wireobinit.cpp:119): trace mess

age: Setting static property map

Thu Sep 18 10:59:05 2008 913548 1 (wireobinit.cpp:126): trace mess

age: CWireObject::InitializeStaticPropertyMap() - reference count: 1

Thu Sep 18 10:59:05 2008 913548 1 Log: level=3 cat=1 msg=35100 p=C

entral Management Server started

Thu Sep 18 10:59:05 2008 913548 1 trace message: SIServerControlle

r::run: Before initialize

Thu Sep 18 10:59:05 2008 913548 1 (sidaemon.cpp:416): trace messag

e: SUNIXDaemon::init: starting

Thu Sep 18 10:59:05 2008 913548 1 trace message: SUNIXDaemon::init

: redirecting stdio to /dev/null

Thu Sep 18 10:59:05 2008 913548 1 (sidaemon.cpp:209): trace messag

e: installing fg terminiation handler

Thu Sep 18 10:59:05 2008 913548 1 (sidaemon.cpp:220): trace messag

e: installing abnormal sys handler

Thu Sep 18 10:59:05 2008 913548 1 (sidaemon.cpp:506): trace messag

e: SUNIXDaemon::init: finished

Thu Sep 18 10:59:12 2008 913548 1 trace message: GetClockSequenceU

pdater: Getting clock sequence updater

Thu Sep 18 10:59:13 2008 913548 1 trace message: SSimpleBuffer::lo

adFromFile: calling mmap(): fileName[/opt/BO/boxir2/bobje/enterprise115/aix_rs60

00/sqlrule.llr], lFileLength47132, fd10

Thu Sep 18 10:59:13 2008 913548 1 trace message: SSimpleBuffer::lo

adFromFile: calling mmap(): fileName[/opt/BO/boxir2/bobje/enterprise115/aix_rs60

00/sqlrule.dfa], lFileLength26824, fd10

Thu Sep 18 10:59:14 2008 913548 1 trace message: Loading: database

subsystem: db2databasesubsystem

This is exactly the same as a normal manual startup of the CMS server except it dies at the loading of the db2databasesubsystem section.

Business Objects is patched to Fix Pack 3.6 and is clustered in a HACMP environment. It is a requirement that when the DB2 database fails, shortly after restarting it can automatically start the Business Objects services remotely.

SSH will work if the command is entered manually from within a secured session, but it is only when automated that the failure occurs. Does anyone have an idea on why this is happening and whether it is in fact possible ?

[denis_konovalov · Accepted solution]

my guess is that when you do it thru the script not all of the env variables are loaded.

Try sourcing .profile and env.sh in your script prior to launching startservers.

[Former Member · Accepted solution]

Good Answer Denis and a distinct possibility which we are investigating as I type. But you know I find it very ironic that in over 3 years working at BO I never once received an answer to a forum post, and as soon as I leave I get a decent response. are you trying to tell me something ?

[denis_konovalov · Accepted solution]

we're going to run into deep offtopic if we continue that line ))

But I believe there is a big push to get the forum format actualy working this time.