Topic: Fast switchover
Colleagues, impart experience, please, who as does DB switching in systems. I.e. experience OLTP of bases under loading is interesting.
Problem essence such - it is necessary to switch on with minimum .
1) you Do preparation for switching ( in , broad gulls to roll away all long transactions if such is)
2) Otstrelit of session, broad gulls once again to be convinced that they were applied.
3) to Push alter database commit to switchover to physical standby with session shutdown...
And here the situation when the command hangs can turn out, thus nobody locks it, in v$session_wait it is possible to see RFS sessions which try to send on any old broad gulls, thus the machine does not rest against a network and in broad gulls not so much. cannot generate end of redo in any way and it turns out that switching cannot be fulfilled, and already goes. Command restart commit to switchover does not help;
Rough restart in restricted with repetition commit to switchover helps.
Who would be desirable to hear as is prepared for switching and that does in similar cases (uses any special event or parameter to rake over the coals to formation process end of redo).
I will specify once again, it not a problem of locks from the user sessions or rollback of transactions, it is a problem at level MRP/RFS/- still.
Basis 12.1 linux_64
As one of variants - reexhibiting dest_x there where you switch.
Thanks in advance for answers!