standby reinstate failed: other options
Date: Thu, 25 Aug 2011 16:30:22 +0300
Message-ID: <OF31ED0EED.33AD46D3-ONC22578F7.0047BF11-C22578F7.004A3187_at_seb.lt>
Hi
Yesterday I've got a situation. Data Guard performed a failover just as advertized. Btw, it proved once and again that active-semiactive environment is preferable to active-passive and alikes. It also proved that the best high avail solution is the one you have the most of knowledge about.
But the problem I got was the failure of reinstate.
In short, reinstate flashbacked the former primary (wanna be standby) and
started managed recovery.
The managed recovery requested a redo sequence which was in the online redo
logs and was never archived. The funny part was that as part of becomming a
new standby the database cleared the online redo logs...
The managed recovery has stuck like that waiting for gap:
ORA-19906: recovery target incarnation changed during recovery
...
Media Recovery Waiting for thread 1 sequence 9652 branch(resetlogs_id)
742441725
Fetching gap sequence in thread 1 branch(resetlogs_id) 742441725, gap seq
9652-9652
The question now is:
is it possible to reinstate the former primary using incremental backups?
Now if I do an incremental backup of new primary - what's use it will have on a flashbacked former primary ?
The tricky parts here are those:
- Resetlogs_change of new primary and the former primary differ. Not sure if incremental backup of different resetlogs can be applied on top of the former primary...
- the former primary was flashbacked to the scn when standby_became_primary as it should. In other words - the former primary got a new incarnation.
- at least the last online(and standby) redo log has gone with the wind - but that's probably not critical
Thank you in advance, Laimis N
Please consider the environment before printing this e-mail
-- http://www.freelists.org/webpage/oracle-lReceived on Thu Aug 25 2011 - 08:30:22 CDT