Tivoli Management Framework Managed Nodes/ Gateways

Goto the Tivoli Home Page
This page details General problems and issues relating to Tivoli Management Framework Managed Nodes and Gateways.

"CreateWindowStation failed: Access is denied"

I've installed a Managed Node on NT, but the OSERV service won't start.  If I look in the oserv log I see the following:

CreateWindowStation failed: Access is denied

Contributed By: Jim Federline, Cliff Hobbs [MVP SMS], Frank Kleinburg,  Nell McCoy
 CreateWindowStation is a Registry write operation so check the following:

  1. Check to see which accounts have access to the '%SYSTEMROOT%' and '%SYSTEMROOT%\System32' directories. By default the 'Everyone' group or at least the 'tmersrvd' account needs Read access to these.  If for security reasons you have removed the 'Everyone' group you can either grant the 'tmersrvd' explicit Read access to this directory or make the 'tmersrvd' account a member of a group that has at least Read access to these directories. (For example if you've removed access to the 'Everyone' group but the 'Users' group has Change access then make the 'tmersrvd' account a member of the Users group).  Now try to start the OSERV service.

    UPDATE 23/1/03 -     Contributed By: Cliff Hobbs [MVP SMS]
    If you're trying to install a Managed Node on an NT Member Server make sure that the LOCAL and not the domain '    tmersrvd' account is either a member of the 'Users' group or has Read access to the     '%SYSTEMROOT%' and '%SYSTEMROOT%\System32' directories otherwise you'll get a '1067' error after the Managed Node has been installed and rebooted whenever you try to start the OSERV service.
     
  2. Grant the 'Everyone' group Read access to the top level HKEY Registry keys.
     
  3. Reboot the machine.
     
  4. Check that the 'Everyone' group or at least the 'tmersrvd' account also has the 'Bypass traverse checking' advanced user right.

As a last resort and only if all else fails make the 'tmersrvd' account a member of the Administrators group, but by doing this you are giving the 'tmersrvd' account more privileges than it needs.
 

"CreateWindowStation failed: The specified module could not be found"

I've just installed a Managed Node on NT, but when I try to start the OSERV service I get the following error message:

CreateWindowStation failed: The specified module could not be found

Contributed By: Nell McCoy, Jacques Mineur
 This error can be caused if the input locales on the machine don't contain the US keyboard.  To check goto Control Panel | Keyboard and select "Input Locales". The Input Locales must contain the US keyboard but it doesn't need to be the active keyboard.

Also make sure that the KBDUS.DLL exists in the %SYSTEMROOT%\System32 directory.
 

Error on wunsub -a "disk_dir create failure"

Contributed By: Doug Sutherland, Ken Matheson and Marc Remes
 If you’re seeing the above error when a Gateway is doing an automatic LCF upgrade via the login policy then here’s a solution… [Go to the article]
 

"FRWII0123E The following errors occurred whilst notifying other applications about the removal of the managed node" error when running "wrmnode" command

One of the Managed Nodes went bang so we rebuilt it.

I ran the 'odadmin odlist' command to get the dispatcher number and then used the 'wrmnode -f <name> -d <id>' command to remove the MN from the database. A couple of 'wchkdb -u' later and it had gone from the database.

So we tried re-installing the MN. It got as far as installing and rebooted but the OSERV wouldn't start with the dreaded 1067 error. After trying all the obvious (KBDUS.DLL, NSLOOKUP both ways, permissions, OS rights, etc.), the OSERV would just not start (the oserv.log had the ' CreateWindowStation failed: Access is denied' message in it), we decided to pull the plug on it and try installing it again.

So, we ran odadmin odlist to get the dispatcher number and used the wrmnode command to delete it from the database. Ran wchkdb -u a couple of times and the database came back clean.

Then we ran 'oinstall -remove' on the MN itself and removed Trip.

I then did a final check with the 'odadmin odlist' command and to my horror the MN is still listed. If I try to remove it using the wrmnode command this is what I get back:

 "FRWII0123E The following errors occurred whilst notifying other applications about the removal of the managed node. The node was removed anyway.

-> FRWSL0007E An authorisation error of type "FRWOG00022E insufficient authorization occurred

Node <name> removed.

Please use wchkdb to clean up any database entries to the remove node(s)"
It now doesn't matter what I try to run, it seems as though there's no way of getting rid of this MN.

Contributed By: Jim Federline
 You're doing fine - your procedure is sufficient, so far. Here's what I follow religiously - the 'odadmin rm_od' command is what you are likely in need of:

  1. wrmnode <managed node>
     
  2. Try wlookup -r ManagedNode <managed node> - does it still exist?
    try wls /Library/ManagedNode | grep <managed node> - is it here?
    try odadmin odlist | grep <managed node> - is the node listed here? get od#
    try odadmin odlist objects <od#> - are more than 3 objects listed?
     
  3. If any of the above is yes, do wrmnode <managed node> -d <od#>
     
  4. Repeat tests in #2
     
  5. If any of the test are true, make sure the oserv is shutdown on the managed node, and run this script (if you're using NT run the script as follows from a Tivoli CLI:
         bash fix_mn.sh <managed_node_name>)
     
  6. Repeat tests in #2 - it should be gone now.
     
  7. Follow with a wchkdb -u.

"FRWSL0007E An authorization error of type "FRWOG0022E insufficient authorization" occurrred" error when running "wadminep <ep> upgrade" from Managed Node

One of the patches I'm applying to my environment includes an upgrade to the Endpoint code.  In the Readme it says to use the 'wadminep <endpoint_name> upgrade' command to remotely upgrade the Endpoint to the latest version.  However when I try to run the command on one of my Managed Nodes/ Gateways I get the following error:

FRWSL0007E An authorization error of type "FRWOG0022E insufficient authorization" occurred

The environment I have is my TMR is installed on NT.  I then have a Managed Node/ Gateway installed on an NT server in a different domain.  There is no trust relationship between the two domains which I think might be causing the problem.  I can upgrade the Endpoints that log into the Gateway I've installed on my TMR server fine - it's only when I try to run the command on the Managed Node on the server in the other domain to upgrade Endpoints that login to this Gateway that I get the error.

Contributed By: Cliff Hobbs [MVP SMS]
 The reason you're seeing the error is because you have a conflict in credentials.  Tivoli knows nothing about this other domain. So you log into your Managed Node/ Gateway using an account Tivoli knows nothing about and so it fails.

To get around this problem setup a new Administrator on your TMR with the same credentials you are using to log into your Managed Node/ Gateway in the other domain.  When you set this new user up, when it comes to Setting the Logins specify the domain name and user ID that you are using in the domain that your Managed Node/ Gateway sits in.

For example your TMR server is called 'TMRSERVER'.  It sits in the 'DOMAINA' domain and you login using the 'TMRADMIN' account.  Your Managed Node/ Gateway on which you are having the problem is called 'SERVER1'.  It sits in the 'DOMAINB' domain and you login using the 'NORMUSER' account.

So when you specify the logins you'll enter 'DOMAINB\NORMUSER' in the 'Add Login Name:' box.
 

How do I Write a ‘select_gateway_policy’?

I have two Gateways. There is NAT between the Gateways and Endpoints. I'd like to write a ‘select_gateway_policy’ to involve the two Gateways but I’m not sure how. How do I configure my TMR in NAT?

Contributed By: Ren Maddox
 I am not quite sure what you are asking, so I will guess that the situation is that you have one or more NATs between your two Gateways and their Endpoints, but no NAT between your Gateways and your TMR Server. With this configuration, it is normally sufficient to configure your ‘select_gateway_policy’ to output the Endpoint-resolvable hostname for each Gateway like so:

echo "$GWOID1|gwhostname1.domain.name"
echo "$GWOID2|gwhostname2.domain.name"

Managed Node Installation Hangs at "Distributing machine independent Client Database"

I've tried installing a 3.7B Managed Node in a remote NT domain.

On the TMR the installation has got as far as saying 'Distributing machine independent Client Database' followed by a single dot (.) and there it sits.

If I look on the Managed Node I'm trying to install everything looks fine - it has installed and the oserv is running.

I'm guessing that because I can't NSLOOKUP from the MN to the TMR, the MN can't tell the TMR it's finished.

Once I get name resolution working, does anyone know if the TMR will update or is there anyway I can tell the TMR the MN is installed without having to delete it and mess around with the 'odadmin odlist rm_od' command?

Contributed By: Cliff Hobbs [MVP SMS]
 Resolving my own problem above was quite easy.

I added an entry to the 'HOSTS' file on the server that I wanted to become a Managed Node. I then used the procedure in the Tivoli Field Guide 'Demystifying the Installation Process' that someone very kindly sent me to remove the Managed Node. Then I re-installed it and everything worked fine.
 

"No Catalog 'TMF' found in the cdrom directory" Error when Installing Managed Node

I'm having an interesting problem when trying to install a Managed Node under 3.7B.

I copied the Framework CD to my server and then installed the TMR using this copy and everything seems okay or at least I thought it was until I tried to install a Managed Node from the Tivoli Desktop.

I specify the Access Method, add the client I want to install as a Managed Node then hit the "Install" button but I get:

Error

No Catalog 'TMF' found in the cdrom directory 'servername path to my Tivoli
install'. You can use the "Select Media..." button to change the media
directory from the gui or pass -c <cd-rom directory> if you are using
command line install.

I did a search for "TMF.IND" and found the file in the following locations:

  • The root of the directory where I installed Tivoli
  • The root of the directory where I installed Tivoli\bin\generic
  • The root of the directory where I installed Tivoli\bin\w32-ix86
  • The root of the directory where I installed Tivoli\lib
  • The root of the directory where I installed Tivoli\lib\w32-ix86\installed

Although the file exists in these locations it is a 0 KB file.  The file also exists in the root of the directory that I copied the Framework CD to and it is 12 KB in size.

If I use the "Select Media..." button and point to the 12 KB file I then get an "Install Options" screen that allows me to specify various directory locations for things such as Libraries, Binaries, Client Database, etc.

This screen also contains three tick boxes at the bottom, two of which are ticked, but the one that's marked "When installing, create 'Specified Directories'" is unchecked.

So now for the questions:

  1. Should I mark the "When installing, create 'Specified Directories'" box or leave it at it's default?
     
  2. Why are the TMF.IND files empty? Did I not install Framework correctly?
     
  3. Should I copy the 12 KB file from the CD copy directory to the root directory of where Tivoli is installed so it always uses this version?
     
  4. Will I need to go through the "Select Media.." process every time I install a Managed Node or once you've done one does Tivoli 'remember' where to find the "TMF.IND" file?

Contributed By: Cliff Hobbs [MVP SMS]
 Seeing as I came across this problem, the following are my comments/ the results of my investigations.  If you have anything to add or any comments please feel free to email me.

  1. You can select the "When installing, create 'Specified Directories'" box.
     
  2. I'm assuming the "TMF.IND" files are 0KB as when you install Framework, it doesn't copy the necessary code over to the TMR server to install the Managed Node which is why you need to use the copy of "TMF.IND" off of the CD (or in this case as the CD has been copied to a directory on the TMR server, point to the "TMF.IND" file in this directory).
     
  3. I've not copied it over and will continue to point the Managed Node installations in the direction of the file in the directory that contains the copy of the CD-ROM image.
     
  4. To be advised ...

"The man_node_skel1 exited unexpectedly" Error on Managed Node

I have a Tivoli environment of a TMR and two MNs. Suddenly today the HPUX MN (also a Gateway), is reporting the error:

the man_node_skel1 exited unexpectedly

The oserv and Gateway on the machine work properly and I can communicate with these EPs. The rest  of Tivoli seems to be okay. Every time I start the oserv process the following information is written to the oserv.log and syslog (hpux:syslog.log)

oserv.log

2002/10/17 14:40:20 -08:
^/other/tivoli/bin/hpux10/TAS/MANAGED_NODE/man_node_skel1 exited unexpectedly, status=5

syslog

Oct 17 14:40:20 L2000 oserv94[11641]: execve failed:
/other/tivoli/bin/hpux10/TAS/MANAGED_NODE/man_node_skel1: errno 13

Contributed By: Ren Maddox
 'status=5' and 'errno=13' both point to permission problems, so that's where I would start my investigation.
 
© FAQShop.com 2003 - 2005

Goto the Tivoli Home Page

 Email the Author