Ok, I was able to see log files after doing:
crsctl set log res "glassfish=5"
and I was able to confirm it was set with:
crsctl get log res glassfish
Get Resource glassfish Log Level: 5
Then, when I try to start, I get:
crsctl start res glassfish
CRS-2672: Attempting to start 'glassfish' on 'my_new_test'
CRS-2674: Start of 'glassfish' on 'my_new_test' failed
CRS-4000: Command Start failed, or completed with errors.
But, glassfish actually *does* start up. It's just the Oracle doesn't know
it:
crsctl status res glassfish
NAME=glassfish
TYPE=ora.local_resource.type
TARGET=ONLINE
STATE=OFFLINE
And in the log file, I see:
more scriptagent_oracle.log
Oracle Database 12c Clusterware Release 12.1.0.1.0 - Production Copyright
1996,
2013 Oracle. All rights reserved.
2016-01-25 14:36:29.588: [ AGFW][414611040] Starting the agent:
/u01/app/12.1
.0/grid/log/my_new_test/agent/crsd/scriptagent_oracle/
2016-01-25 14:36:29.589: [ AGENT][414611040] Agent framework initialized,
Proc
ess Id = 32021
2016-01-25 14:36:29.589: [ USRTHRD][414611040] Failed to initialize agent,
could
not get type name..
2016-01-25 14:36:29.589: [ AGFW][414611040] Agent is exiting with exit
code:
-1
I *think* the most relevant error is "Failed to initialize agent, could
not get type name..", but I have no idea what it means, and searching MOS
and Google has not been successful, thous far.
Any thoughts, ideas, or suggestions would be appreciated.
Thanks,
-Mark
On Mon, Jan 25, 2016 at 6:49 PM, Ls Cheng <exriscer@xxxxxxxxx> wrote:
Hi
In 12c you have to look in $ORACLE_BASE/diag/..................
(DIAGNOSTIC_DEST)
On Mon, Jan 25, 2016 at 8:02 PM, Mark J. Bobak <mark@xxxxxxxxx> wrote:
Hi Ilmar,
Thanks for the reply. It appears the log doesn't exist in 12c, or it has
a different name....
[oracle@my_new_test log]$ pwd
/u01/app/12.1.0/grid/log
[oracle@my_new_test log]$ find . -name script\*
No results under DRS_HOME/log for any filename starting with 'script'.
I'll poke around, see if I can figrue out the correct log file.
-Mark
On Mon, Jan 25, 2016 at 11:45 AM, Ilmar Kerm <ilmar.kerm@xxxxxxxxx>
wrote:
Hi
We have run it successfully for years to provide high availability for
MySQL and currently I'm also setting it up to have a cold standby for
Oracle database in Windows.
I haven't seen the attributes START_PROGRAM,STOP_PROGRAM and PID_FILES.
Are they new in 12.1?
In 11.2 I've only used attribute ACTION_SCRIPT that points to the script
that handles startup/stop and check actions.
Could you have problem with the check action? After start and stop
actions clusterware will also invoke check action to verify that
startup/shutdown was successful.
You could verify it from scriptagent logfile:
$CRS_HOME/log/<hostname>/agent/crsd/scriptagent_<username>
It logs all the actions executed and the responses.
My old presentation about this topic:
http://www.slideshare.net/ilmarkerm/making-mysql-highly-available-using-oracle-grid-infrastructure
Ilmar
On Mon, Jan 25, 2016 at 5:28 PM, Mark J. Bobak <mark@xxxxxxxxx> wrote:
Hi all,
Has anyone sucessfully done this? In my case, I've managed to figure
out how to create teh necessary resource, but it doesn't actually work.
What I'm tying to do, is start up a Glassfish server (running on the DB
server, single instance, in my case), with a dependency on the database. I
have ORDS (formerly known as APEX Liostener) deployed in Glassfish, and if
I just start Glassfish via init.d, it will start before the database, and
gets stuck reporting 404, even after the db starts, until I bounce
Glassfish. So, I want to start if from GI home, using crs/has, and have a
dependency on the db, to solve the startup sequence issues. (Also, to make
it (presumably?) stop and stat automatically whenever the DB is bounced.)
So, I created the resource, like so:
crsctl add resource glassfish -type generic_application -attr
"START_PROGRAM='/etc/init.d/glassfish_domain1
start',STOP_PROGRAM='/etc/init.d/glassfish_domain1
stop',START_DEPENDENCIES='hard(ora.orcl.db)',STOP_DEPENDENCIES='hard(ora.orcl.db)',PID_FILES=/var/run/glassfish.pid"
This seems to have worked, no error is returned, and one can query the
resource and see it exists:
crsctl status res glassfish
NAME=glassfish
TYPE=generic_application
TARGET=OFFLINE
STATE=OFFLINE
But, when I try to start it, I get this:
crsctl start res glassfish
CRS-2672: Attempting to start 'glassfish' on 'my_new_test'
CRS-2674: Start of 'glassfish' on 'my_new_test' failed
CRS-2679: Attempting to clean 'glassfish' on 'my_new_test'
CRS-2681: Clean of 'glassfish' on 'my_new_test' succeeded
CRS-4000: Command Start failed, or completed with errors.
And, at the end, the glassfish server is, in fact, running. So, it
thinks it failed to start, but it starts, then it does the clean, and it
thinks it cleaned everything up (remove logifle and pid file and kill
process) but it in fact remains running.
I suspect that there is a 'tamplate' for the startup/shutdown script?
I found a doc on MOS, that says my script must accept
'start','stop','clean, and 'check' as parameters. When I run my sript from
the command line, it works beautifully, and those parameters all work as
expected.
So, my next guess is return values? Not sure what to try next...
Has anyone gotten this to work? Any idea what I'm missing from my
script?
Thanks,
-Mark
--
Ilmar Kerm
