Re: Q: farsync questions
- From: amihay gonen <agonenil@xxxxxxxxx>
- To: Ludovico Caldara <ludovico.caldara@xxxxxxxxx>
- Date: Tue, 14 Oct 2014 13:44:01 +0300
I want to simulate a situation where the FS has data not shipped to standby
and primary is gone.
now I've saw error complaining that the primary is not accessible ,that ok.
But my simluation has a "twist" where the FS is also crashed .
I don't understand although why the FS was looking for primary redo .
btw ,I'm working in Max Availability (since I first "kill" the primary"
and then the FS )
On Tue, Oct 14, 2014 at 12:38 PM, Ludovico Caldara <
ludovico.caldara@xxxxxxxxx> wrote:
> Hi Amihay,
> the FS will have errors since it has no connection to the primary. How can
> the FS know if the primary is not alive and generating new redo? That's the
> situation when you analyze the situation and take the risk of activating
> the failover...
> Having sync transport does not mean having complete recovery, the best
> protection mode that you can achieve is Max Availability if you're using
> far sync instances...
>
> If you look at the sequence# and block# of your RFS processes (or standby
> log files)
> you should have the last transactions and have those transactions sent to
> the standby up to the same sequence/block. have you verified this?
> you can check v$managed_standby view on the alive instances.
>
> Best regards
> --
> Ludo
>
>
>
> 2014-10-14 11:19 GMT+02:00 amihay gonen <agonenil@xxxxxxxxx>:
>
>> Hi all , I've started playing with farsync to learn better how it work.
>>
>> I've setup primary and farsync on the same instance (I know it is not
>> best practice , but for learning propuse). I've create a standby on another
>> machine and verified everything is working as expected.
>>
>> Then I've stop the standby (simluate network outage) , put some data in
>> primary db + commit (no switch log ) and crash both primary db and farsync
>> instance (in this order).
>> After that , I've renamed the directory contains the primary db datafiles
>> and redolog (simluate database loss)
>>
>> When starting the FS instance , i got errors in the alert log of the FS
>> complaining it doesn't find the redologs of the primary !! .
>>
>> Any idea why ? it seems the FS read the primary redo log to complete
>> recovery? this is strange to me since it is sync transport
>>
>>
>
Other related posts:
