Last week we removed the configuration from a test system, where we didn’t had the luxury to scrap it. And how hard could it be? Just disable the transport of the log-files, no force logging and replace the log_archive strings with uhm..nothing.
Until one node decided to go down for some obscure reason. No worries, we thought..It’s a RAC, load balancing kicked in, and we could keep on working. However, when trying to restart the instance we got a bit of a surprise: It just wouldn’t start!
In the alertlog/trace file we could find the reason:
ORA-16188: LOG_ARCHIVE_CONFIG settings inconsistent with previously started instance LGWR: terminating instance due to error 16188 and instance will be terminated.
Huh? How can that be? We are using a spfile and we used the command:
sql> alter system set log_archive_config='' scope=both sid='*';
Well..you must guess it already…we ran into a bug. No worries..It’s all solved in Oracle 12.1c…*sigh*. The problem is this: “The v$dataguard_config is not updated between instances”.
Now we now the cause..let’s fix it! Since we have a remaining node (and theory says all the nodes in a RAC will either go down, or at least won’t start up, so one need to be carefull when removing ADG configurations) we log into the surviving node and issue the following command:
sql> alter system set log_archive_config=NODG_CONFIG scope=both sid='*'; Database altered.
Now go to the downed node, and try to start it up with the same spfile/pfile (so no changes need to be done on this node!).
Luckily: it worked for us, so this story ends here.
For more Oracle tech details, check the link:
I hope this information helps out when in a pinch. Or when not in a pinch. Which is even better.. 😉
‘Till next time!