I am evaluating TSAM for our client and I have problem monitoring Tuxedo 9.1 domains with TSAM 11gR1:
with Tuxedo monitoring policy enabled I saw no data collected in Tuxedo metrics.
To investigate the issue I set up two simple domains in a test environment:
- one based on tuxedo 11gR1 (with TOUPPER service)
- second on 9.1 RP 110 (with TOLOWER service).
These domains are connected, services are exported to each another.
LMS server is defined in both, the only difference is URL of dataserver.
In Tuxedo panel of TSAM console both domains are discovered with their machines, servers and services but 9.1 domain is displayed with red icon, it's model is null and status UNK.
There are no errors from LMS in ULOGs.
I defined identical policies for each domain (only BRIDGE and GWWS are disabled) to trace call paths and here is the problem:
- I see no data collected for calls made in tuxedo 9.1 domain locally.
- When I call TOUPPER from client via 9.1 domain I see only part of call path
(from GWTDOMAIN in 11gR1 domain, application server and back to GWTDOMAIN, rows for client and GWTDOMAIN in 9.1 are missing).
- When I call TOLOWER service from Tuxedo 11gR1 domain I see whole call path
(ie. from client in 11gR1 through GWTDOMAIN in 11gR1, GWTDOMAIN in 9.1, application server in 9.1
and the way back).
- Call TOUPPER locally in 11gR1 domain is obviously visible in console.
Is my configuration fully supported?
If yes, what could have I done wrong? I will provide further details if needed.
If you are monitoring a Tuxedo 9.1 domain, you are using TSAM 1.1. TSAM 11gR1 cannot push monitoring policies out to a TSAM 1.1 system. You will have to define the monitoring policies on the Tuxedo 9.1 itself through environment variable, tmadmin changemonitor command, or MIB requests.
Once you define the monitoring policies on the Tuxedo 9.1 system, TSAM 11gR1's console will show the collected data.
Oracle Tuxedo Chief Architect
Todd, thank you for kind response, bu I need more assistance, I'm afraid.
Setting TMMONITOR (TMMONITOR=app,svc,tran,sys::) in domain's environment on Solaris solved the problem,
data is collected now and can be seen in console.
There is still an issue on HP-UX: the setup is the same as on Solaris (Tuxedo 9.1 RP 99 instead of RP 110) but the differences are minor I think.
When a service is invoked from another domain or locally "LIBTUX_CAT:6623: ERROR: TMMONITOR failed to invoke the monitoring plug-in" is logged
by application serwer as well as by GWTDOMAIN.
I suspected libmonplugin.sl, becaused removing equivalent library on Solaris caused the same symptom, but this library is loaded by application server,
I even forced it by setting LD_PRELOAD and checked with pldd.
Setting TMMONITOR to different values does not help.
What is the plug-in which can't be invoked?
What am I missing?
I can't think of anything in particular that could be causing that problem. What version of HPUX and platform are running Tuxedo on? If you could post the output from the tmadmin initial display and the output of uname -a (or whatever the equivalent is on HPUX) that would be helpful. I suspect that you are going to need to open a support case with Oracle Customer Support, but let's see what we can do in the meantime.
Oracle Tuxedo Chief Architect