IBM Director 4.12 won't start

[Shinanygnz]

Hi there

We've recently had a problem with our IBM Director installation. It was running on v4.1 quite happily but then just stopped. On restarting, it starts the "IBM Director Support Program" service OK, but the "IBM Director Server" service starts and then fails. I have tried reinstalling v4.1, then uninstalling it and installing v4.12 but the problem remains. The error reported in the event log is as follows,
TWGServer
Java Exception
Event ID 100
Unhandled JVM Exception
Application: 'TWGServer (TWGVersion 4.12 2004-02-22 build id 464)'
Thread Name: TWGServer
Thread Group: ServiceThreadGroup
Exception Type: java.lang.OutOfMemoryError
Stack Trace: java.lang.OutOfMemoryError

The system is running on a server with 2Gb RAM, W2K sp4 and all hotfixes. Database is MS SQL2000 (sp3) on the same server.

IBM's web site is useless (as ever) and Director is not supported by WESS. Anyone got ideas?

Cheers

Steve

 

[Dimarc67]

How many clients do you manage with this particular Director server? If it's less than 1000, I'd let Director use Windows' built-in Jet database engine instead of installing SQL.

Even just for testing purposes, try installing Director without SQL and see if that gets you any further...

David Marcus
Network Administrator
CurtCo Publishing
DavidM@CurtCo.com

 

[catorze]

The error you referenced has been seen in IBM Director v4.11, but there were numerous java fixes in v4.12. Although the Build date is correct for v4.12, the build ID cited in the error does not appear to be correct. Take a look at your version.key from the Director install directory to see what it shows. You can also see where the Director server is crashing by double-clicking the IBM Director icon in the system tray and watching the services load. If it is dying on the database initialization, and you've confirmed the SQL server is running and accessible, you can run CFGDB.EXE to confirm the ODBC connections to the SQL database. If that doesn't work, you can also use the command to change to a JET database to rule out any SQL-Director communication/connectivity problems.

You lost me on IBM's site being useless and WESS supporting IBM Director - if you go to the right place on IBM's site:

http://www.pc.ibm.com/support

and select "Systems Management" from the lower left section of the right pane, you can get what you need for IBM Director. Otherwise, IBM Director is supported by IBM. It's fee-based, but you can get to the right folks by calling (800)IBM SERV and selecting Software Support.

 

[Shinanygnz]

Thanks for those, didn't know about the icon info. Tried JET, but the same problem happened. In the end I managed to get it working again after uninstalling (again), ripping all the info out of the registry and reinstalling. I restored the SQL database from the last day we know it was definitely working and looking in SQL the tables contain information. However, whenever the services start up, it wipes this out! I've searched IBM's website and the redbooks site and Google too, but can't find any info on restoring the Director database. Any idea how to stop it doing this? Are there config files we need and if so, where and which are they (again the Director documentation doesn't mention any)?

In reference to IBM's website being useless, try searching for something (like IBM Director SQL restore). You get loads of irrelevant rubbish and can never find what you want, assuming it even exists.
We emailed WESS to find out about getting support for Director and got a short "this product is not supported" reply.

 

[catorze]

If Director's persistent data gets pooched, rather than rip & reinstall, you can run TWGRESET.EXE A word of caution, this restores Director to a sterile installation, so any customizations you made (EAP, User Config, etc) are gone. If you had a backup of your Director config (TWGSAVE.EXE), you could run TWGRESTORE to put back the working stuff. Hint: After Director is completely configured (Relay Agents, Event Action Plans, User Administration, etc) it's a very good idea to run TWGSAVE.EXE for just such an emergency. If it's connectivity to the database that gets pooched, you can run CFGDB.EXE to configure/restore/add the database ODBC entries.

 

[Shinanygnz]

No, we didn't have a copy of the config as we'd never heard of twgsave.exe
I've run it now, but the horse is unfortunately long gone. When I tried a restore from backup tape of the previous install and it's config, Director crashes on start up. Oh well.

Got a link to a command reference for Director by any chance?

Thanks for the help.

 

[catorze]

The only availble CLI IBM Director currently supports is for the MPA Tool (ASM Processors). The one I used is in my head.

 

[margudmju]

Dear Catorze
What do you mean by "Relay Agents"? It was my assumption that this is not a supported function in IBM Director. Have you found a way around it or do you have a suitable solution ?

Best regards,
Martin

 

[catorze]

By default, IBM Director server can discover and manage Agents in the same subnet as the Director server. To discover Agents in remote subnets, and since most IT police will not forward subnet directed broadcasts across a router, you should configure Relay Agents (from the Director server Discovery Preferences) to receive Director Server discovery packets and broadcast them in their local subnet. That way communication across the router is point-to-point, and the broadcast is limited to within subnets only. This is not only a supported function, it is a necessary IBM Director configuration to effectively manage beyond the Director Server's subnet.

 

[Stiddy]

Hey everyone. I am having a similar problem. Director seems to run well for about a week or so then I get this error in the eventlog. This has happen on every version and I am getting tried of rebuilding. Thanks 4 any help.

Unhandled JVM Exception
Application: 'TWGServer (TWGVersion 4.22 2005-05-26 build id 574)'
Thread Name: ServiceNode <TWGServer>
Thread Group: ServiceThreadGroup
Exception Type: java.lang.NegativeArraySizeException:
Stack Trace: java.lang.NegativeArraySizeException:
at com.tivoli.twg.libs.Command.AddOutputParm(Command.java:454)
at com.tivoli.twg.libs.Command.UseFlatOutputBuffer(Command.java:941)
at com.tivoli.twg.libs.ServiceNode.InternalDataReceive(ServiceNode.java:1592)
at com.tivoli.twg.libs.ServiceNode.ProcessCommands(ServiceNode.java:703)
at com.tivoli.twg.libs.ServiceNode.run(ServiceNode.java:1925)
at java.lang.Thread.run(Thread.java:513)


HTH - Stiddy

 

[Dimarc67]

Was this solved? Have the identical event which showed up spontaneously today on a server that's been running IBM Director without issue for months now...

 

[bbird74]

maybe reinstalling the JRE?

 

[Dimarc67]

IBM Director has its own JRE that is installed within its Program Files directories. No method that I'm aware of to re-install other than re-installing all of Director.

As it happens, I had JRE 1.5.0 installed for unrelated needs on the server, and re-installing it had no effect.

Attempted a Director re-install, and answered "no" to deleting the data files. Director continued to fail with the identical error event.

Finally performed full uninstall, including the data files, and reinstalled. Director services started successfully. Basically, for me this narrows it to some sort of data file corruption issue. Mine is a relatively small organization--1 server, 50+ clients--so reinstallation and rediscovery of data is practical, but what if I'd had 500 clients and 50 servers...