You should set the trap in sqlplus, not dgmgrl. The trace file will be
written by Oracle kernel, when the error happens.
Regards
On 01/20/2017 09:04 PM, Andrew Kerber wrote:
this works in dgmgrl?
Sent from my iPhone
On Jan 20, 2017, at 7:55 PM, Mladen Gogala <gogala.mladen@xxxxxxxxx <mailto:gogala.mladen@xxxxxxxxx>> wrote:
Hi Andrew,
Why not trap the error:
ALTER SYSTEM SET EVENTS='48189 TRACE NAME ERRORSTACK FOREVER, LEVEL 16';
When the error occurs, it will emit a trace file which will, with any luck, contain the information you need.
Regards
On 01/20/2017 05:50 PM, Andrew Kerber wrote:
So here is where I am. On the primary side, lets call it testp, switchover works fine using dgmgrl. (switchover to testd).
I go over to the testd side, and run a switchover command in testd (switchover to testp), I get the error message:
Error: ORA-48189: OS command to create directory failed
Error: ORA-16625: cannot reach database "TESTD"
However, if I go back over to the original primary side, and run the same switchover command there, it works fine.
I am sure its some sort of privilege problem, but even with the trace level set to support, it wont tell me what command is failing. And unfortunately, I cant even tell from the error message whether its on the testp or testd side that it is trying to create a file, though it always seem to be when its communicating to testd.
There is no command in the drc*.log or the alert log. Is there anywhere else I could look?
On Fri, Jan 20, 2017 at 3:38 PM, Andrew Kerber <andrew.kerber@xxxxxxxxx <mailto:andrew.kerber@xxxxxxxxx>> wrote:
I tried that one already. It simply wont show the command that
is failing:
Properties:
FastStartFailoverThreshold = '30'
OperationTimeout = '30'
TraceLevel = 'SUPPORT'
FastStartFailoverLagLimit = '30'
CommunicationTimeout = '180'
ObserverReconnect = '0'
FastStartFailoverAutoReinstate = 'TRUE'
FastStartFailoverPmyShutdown = 'TRUE'
BystandersFollowRoleChange = 'ALL'
ObserverOverride = 'FALSE'
ExternalDestination1 = ''
ExternalDestination2 = ''
PrimaryLostWriteAction = 'CONTINUE'
On Fri, Jan 20, 2017 at 2:47 PM, Bernhard de Cock Buning -
GRID-IT <bdcbuning@xxxxxxxxxx <mailto:bdcbuning@xxxxxxxxxx>> wrote:
Hi Andrew,
Would expect to see more details in the drc logfile.
If you find out it is related to one server(s) cluster, It
starts to look like it is related to the log/trace directories.
Have a look at the ADR on that host and the permissions,
maybe it is related to the diag_dest parameter on the
standby location where it defines an incorrect location. But
I would in that case expect you would already have issues
for some time and not only when you perform a switchover.
But if you need more tracing info you can execute in dgmgrll
(assume you use 11.2.0.3 or higher)
Admin is the default tracelevel, change it to support.
edit configuration set property tracelevel=support
Hope this helps
Greetings,
Bernhard
Op 20 jan. 2017, om 21:32 heeft Andrew Kerber
<andrew.kerber@xxxxxxxxx> het volgende geschreven:
interestingly, if I moved over to the original primary
servers to run the command, the switchover succeeds. It
only fails with the error on one cluster, which means I at
least know for sure what servers the error is coming from. Now if I could just figure out what command.
On Fri, Jan 20, 2017 at 2:10 PM, Andrew Kerber
<andrew.kerber@xxxxxxxxx> wrote:
I was trailing that file, and it does not contain the
command that failed. Here is what is in the broker log
file:
Failed to connect to remote database TEST. Error is
ORA-48189
Failed to send message to site TEST. Error code is
ORA-48189.
database TEST unable to contact primary database for
version check; status ORA-48189
completing bootstrap of this database
Notifying Oracle Clusterware to buildup
On Fri, Jan 20, 2017 at 2:05 PM, Bernhard de Cock
Buning - GRID-IT <bdcbuning@xxxxxxxxxx> wrote:
Hi Andrew,
Check out the broker logfile, when correct this
will explain where it goes wrong.
You can find this logfile drc*.log in the same
location as the alert.log of the database.
Be sure you check both location (the primary and
standby)
You can also enable additional tracing in the
broker but as mentioned I believe the drc log will
contain the error.
Greetings,
Bernhard
Op 20 jan. 2017, om 20:56 heeft Andrew Kerber
<andrew.kerber@xxxxxxxxx> het volgende geschreven:
I am getting a strange error when I am trying to
switchover to my standby database. The first
switchover worked fine, so I am switching over to
my original primary, and I am getting this message:
DGMGRL> switchover to 'TEST';
Performing switchover NOW, please wait...
Operation requires a connection to instance
"TEST3" on database "TEST"
Connecting to instance "TEST3"...
Connected as SYSDBA.
Error: ORA-48189: OS command to create directory
failed
Error: ORA-16625: cannot reach database "TEST"
It looks like it is trying to create a directory,
but I cannot figure out what directory it is
trying to creat. I expect its a simple ownership
problem, but I cannot figure out what directory I
need to work on. I tried setting the trace level
to support in dgmgrl, but it didnt show the
information I needed. Any ideas?
-- Andrew W. Kerber
'If at first you dont succeed, dont take up
skydiving.'
-- Andrew W. Kerber
'If at first you dont succeed, dont take up skydiving.'
-- Andrew W. Kerber
'If at first you dont succeed, dont take up skydiving.'
-- Andrew W. Kerber
'If at first you dont succeed, dont take up skydiving.'
--
Andrew W. Kerber
'If at first you dont succeed, dont take up skydiving.'
--
Mladen Gogala
Oracle DBA
Tel: (347) 321-1217
