Act Exception starting AppServers with wscp

[Einstein47]

I'm scratching my head with this one. I have several servers running WAS 4.02. I am using wscp to stop the nodes before shutdown and daily to resync with the database. Then after the node is stopped, I restart the admin server and then have to manually start the AppServers with wscp. (are you with me so far)

For all but one of my servers (AIX 4.3.3 ML9), this works well. However, on my QA server, I will often get a message in my output log saying,

Starting AppServer /Node:aixqa/ApplicationServer:Default Server/...
WSCP0002E: An "Act exception" was received; exception information: null

This hasn't happened on any of my other servers. The wscp script is VERY simple:

foreach ejbserver [ApplicationServer list] {
  puts "Starting AppServer $ejbserver..."
  ApplicationServer start $ejbserver
  puts "   ... started."
}

The odd thing is that this error only happens when I run the script from cron. I have never encountered this error when I run the script from the command-line. Also, it isn't consistent. About 50% of the time it will run as it should. So the mystery gets darker.

Where would I even look to resolve this? Any help would be appreciated. Einstein47
("For every expert, there is an equal and opposite expert." - Arthur C. Clarke)

 

[fearo]

Hi,

WebSphere provides a tcl file which can be used to output more information about what wscp is doing. The file is called init.tcl and should be installed under /usr/HTTPServer/htdocs/en_US/WSsamples/tcl
Unfortunately I have never it before so I cannot advise you. Check the infocenter for more details.

Is WebSphere running as the same userid as the cron job?

Is the AdminServer definitely started and "open for ebusiness" before trying to start the AppServer? Is the AppServer being started first in the list?

I don't think that this is specifically a WAS problem but more likely an environment problem at the time of running the script.

Sorry!

Eddie

 

[Einstein47]

I think I have found the problem. We do a nightly backup of the local Informix database on the development servers with TSM. This means the database was down for most of the night. If the database backup still hadn't finished when the cron job restarted WebSphere, then the JDBC driver would freak and the AppServers wouldn't start.

The Nanny process would keep trying to start the Admin Server and when that finally came up (just after the Informix database completed its backup and was restarted). Then we could start the AppServers without fail.

Whew!

Now we start the Informix DB backup earlier so that it finishes before the websphere restart. I really don't know if cycling WebSphere is necessary, but it was back in version 3.5 and we have just kept those processes in place. (I work for the government - if it ain't broke, keep doing it until it gets that way.)

I include this info, in case anyone else has the same issue. Einstein47
("For every expert, there is an equal and opposite expert." - Arthur C. Clarke)