swinstall using wrong domain

[Robin Sharpe]

Hi,
I'm trying to setup SD-UX so I can push a software package to many hosts from a central depot. To do that, I am running the following on a remote host (wau003):

swinstall -s wau026:/var/opt/mx/depot11 AgentConfig.SD-CONFIG

It contacts the depot on wau026, and the analysis succeeds. In the execution phase, it fails with the following in the swagent.log:

* Beginning the Configure Execution Phase.
* Setting SD host and root ACLs for root@wau026.berlex.com ...
WARNING: An attempt to get the network host entry for
"wau026.berlex.com" failed. This may result in denial of
access to users and agents at this host. Check the spelling
of this name, then your "/etc/hosts" file, or your
"/etc/resolv.conf" file and DNS resolver configuration. The
nslookup program may be helpful in isolating this problem.


"wau026.berlex.com" is no longer valid... that is our old domain name. It is not in /etc/resolv.conf. /etc/hosts and DNS are working correctly. Why is swinstall using the berlex.com domain, and how can I reconfigure it?

I have already verified the entries in the /var/adm/sw/security files and restarted swagentd on both machines. I can swlist software in the central depot from the remote host. Appreciate any help, as this really has me puzzled...

[A. Clay Stephenson]

My best guess is that you need to run swreg -u on wau026 to unregister your depot and then run swreg again to register the depot with the new domain name. I would then cd to /sbin/init.d and execute ./swagentd stop and ./swagentd start on both boxes. It would also be wise to run nslookup from both boxes.

[Robin Sharpe]

Thanks Clay, but still same result.

It seems to me the problem is on the remote machine (wau003)... why is it trying to access wau026.berlex.com? It must be getting the domain from some config file somewhere... the domain is long gone (months) and all machines were recently rebooted during a mass IP change project.

I did have some "berlex.com" references in the files in /var/adm/sw/security (on wau003), but I removed them and left the host name unqualified, and restarted swagentd. I don't have to reboot again do I? In the meantime. I'll look for another remote machine that does not have any berlex.com references.

[Robin Sharpe]

OK, an update... I tried the same thing from another machine that had no trace of berlex.com, and got the same result.

I also tried using wau003 as the central depot (it also has the mx depot local) from theis third machine (wau032), and it looked for wau003.berlex.com.

As a last ditch effort, I tried using IP address of wau026 (man page says you can do that).... it STILL looks for wau026.berlex.com!!!

Sure wish I could find this "ghost in the machine"...

--Haunted in NJ--

[Michael Steele_2]

Dear Haunted:

On the depot server that you are 'pulling' from:

ls -la /var/adm/sw/.sdkey (* there?
touch /var/adm/sw/.sdkey (* if not there
setaccess client_server
swacl -l root

[John Payne_2]

Robin,

Do you have a /{root home directory}/.sw/defaults.hosts file on wau003? Does this file have the wau026.berlex.com entry in it (for the swinstall.hosts_with_depots entry)? If so, you can wither change this to the correct entry, or remove that line, or remove the entire file. You can edit this file, and try the install again for the machine without having to restart anything.

Hope it helps

John

[Ermin Borovac]

Can you please check files in /var/opt/mx/depot11/catalog/dfiles on wau026?

INDEX INFO _ACL _LOCK_FILE _OWNER _PROD_DFLT_ACL

See if wau026.berlex.com is still in there ...

[Robin Sharpe]

Thanks for all the replies, I have tried all of your suggestions... still no luck. Here is what I found though:

.sdkey is there

I'm not familiar with the setaccess command - I found it, but "client_server" was interpreted as a host name (it was trying to resolve it and failed of course). In this scenario, wau003 is the client and wau026 is the server.... what is the exact syntax for setaccess?

wau026:/home/root> swacl -l root
#
# swacl Installed Software Access Control List
#
# For host: wau026:/
#
# Date: Sat Dec 2 18:27:18 2006
#

# Object Ownership: User= root
# Group=sys
# Realm=WAU026.us.schering.net
#
# default_realm=WAU026.us.schering.net
object_owner:crwit
user:root@localhost:crwit
any_other:-r---
wau026:/home/root>


There were some strange entries in the default_hosts file on wau003... wau003 and wau026 were repeated many times, and there were a few wau003.berlex.com, but no wau026.berlex.com. I removed the file, but got the same error.

The files in the mx/depot11 directory did not have any berlex.com, but they had loopback... is that right? Here are the contents of some of them:

wau026:/var/opt/mx/depot11/catalog/dfiles> more _ACL
# default_realm=loopback
user_obj:rwict
any_other:r----
wau026:/var/opt/mx/depot11/catalog/dfiles> more _LOCK_FILE
wau026:/var/opt/mx/depot11/catalog/dfiles> more _OWNER
# default_realm=localhost
num_entries=2
user:root:00
group:sys:00
wau026:/var/opt/mx/depot11/catalog/dfiles> more _PROD_DFLT_ACL
# default_realm=unknown
user_obj:rwict
any_other:r----
wau026:/var/opt/mx/depot11/catalog/dfiles>

-"Still Spooked"

[Ermin Borovac · Accepted solution]

When you swinstall AgentConfig.SD-CONFIG fileset as a part of configure phase it runs the following script.

/var/opt/mx/depot11/catalog/AgentConfig/SD-CONFIG/configure

So check this file on wau026; you may find the following line

/usr/lbin/sw/setaccess wau026.berlex.com

You can try running the following command to recreate AgentConfig depot.

mxsetup -D

[Robin Sharpe]

Ermin,

That appears to be the source of the berlex.com! Thanks for solving the mystery... 10 points for you!

But, now I am getting errors in the mxsetup about "no such file":

======= 12/04/06 08:54:02 EST BEGIN swpackage SESSION

* Session started for user "root@WAU026.us.schering.net".

* Source: wau026:AgentConfig.psf
* Target: wau026:/var/opt/mx/depot10
* Software selections:
AgentConfig,a=HP-UX_B.10.20_700/800


* Beginning Selection Phase.
* Reading the Product Specification File (PSF)
"AgentConfig.psf".
* Reading the product "AgentConfig" at line 27.
* Reading the fileset "MX-CONFIG" at line 45.
ERROR: Cannot access the source file
"/etc/opt/mx/config/.bconfig.cms": No such file or directory
(2).
ERROR: Invalid definition for the fileset "MX-CONFIG", beginning at
line 44. Details described above. The error(s) must be
corrected before the parsing can continue.
ERROR: Invalid definition for the product "AgentConfig", beginning at
line 27. Details described above. The error(s) must be
corrected before the parsing can continue.

======= 12/04/06 08:54:02 EST END swpackage SESSION


I got this for 11.0 and 10.20... I wonder if the best approach now would be a total reinstall of "mx"... How would that be done, Service Control Manager? What exactly is "mx" and where is it documented?

[Robin Sharpe]

OK, re-installing ServiceControl Manager looks like more work than I want... I did a little digging, and figured out that all I really need is to run the setaccess script on each client --

/usr/lbin/sw/setaccess wau026 \*

After doing that, SD is allowing me to add the corresponding client as a target, which is my goal so I can remotely install software. I've run this on several clients, but two of them are just hanging at the first swacl command:

swacl -l host -M user:root@wau026.us.schering.net:a

They even hang just doing this:

wau028:/var/adm/sw> swacl -l host
#
# swacl Host Access Control List
#
# For host: wau028
#
# Date: Mon Dec 4 11:35:47 2006
#


Any ideas what would cause such a hang? I don't see any entries in any of the sw log files...

Regards
Robin

[Robin Sharpe]

Never mind... did a little more research and found a tip about restarting swagentd, which I did, and which worked on both clients. Now I think I'm in good shape for software distribution. Thanks again to all who have helped me out!

Regards
Robin