This mark in the controlfile is like a preventive measure against inaccurate actions during incomplete recovery. Indeed, it is much easier to make a mistake due to variety of possibilities. And it is actually a very rare occasion when only controlfiles are damaged, while the rest of the database is ok. And it is even more rare, when the redo stream is reapplied completely up to the last existing redo record.
Let me remind you that 'USING BACKUP CONTROLFILE' actually means - 'go farther along the redo stream than it is recorded in controlfile'. So you know, that there is more history in the life of this database, and oracle - doesn't. It takes all the info from control, redo and datafile headers and these are the only sources of information available to oracle. So oracle believes you and applies logs with redo sequence numbers which are higher than recorded in controlfile. But oracle's belief is not blind and complete. So it/he/she puts a mark about what is going on into controlfile. Just in case you'll restart the instance ;-).
What IS really surprising - this mark is not reset when the real end of redo stream is encountered (in the last available online redo). Well, I treat it as a rule. Anyway, there are only two possibilities - reset the mark or not. Oracle is well known for its "lazy" approach in many areas (delayed logging block cleanounts, lazy global lock release to name a few).
HTH - Andrey
Alessandro Vercelli wrote: >> No, it is a consequence of using 'incomplete recovery' clause of RECOVER command, i.e. any of UNTIL and/or USING BACKUP CONTROLFILE. >> When you do so, a mark is placed in the controlfile used during RECOVER command execution, which means - "Insist on opening database with RESETLOGS with this particular controlfile". >> > > ....I accept your description, but it's strange as, during recovery, I used "recover database using backup controlfile until cancel;" > > Thanks, > > Alessandro
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=KOI8-R" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> <tt><br> ??? Alessandro,<br> <br> This mark in the controlfile is like a preventive measure against inaccurate actions during incomplete recovery. Indeed, it is much easier to make a mistake due to variety of possibilities. And it is actually a very rare occasion when only controlfiles are damaged, while the rest of the database is ok. And it is even more rare, when the redo stream is reapplied completely up to the last existing redo record.<br> <br> Let me remind you that 'USING BACKUP CONTROLFILE' actually means - 'go farther along the redo stream than it is recorded in controlfile'. So you know, that there is more history in the life of this database, and oracle - doesn't. It takes all the info from control, redo and datafile headers and these are the only sources of information available to oracle. So oracle believes you and applies logs with redo sequence numbers which are higher than recorded in controlfile. But oracle's belief is not blind and complete. So it/he/she puts a mark about what is going on into controlfile. Just in case you'll restart the instance ;-).<br> <br> What IS really surprising - this mark is not reset when the real end of redo stream is encountered (in the last available online redo). Well, I treat it as a rule. Anyway, there are only two possibilities - reset the mark or not. Oracle is well known for its "lazy" approach in many areas (delayed logging block cleanounts, lazy global lock release to name a few).<br> <br> </tt>HTH<br> - Andrey<br> <br> Alessandro Vercelli wrote: <blockquote cite="midIYBV2K$3ECB1E5DE31F5EA436D1E6C48D14A480@(protected)" type="cite"> <blockquote type="cite"> <pre wrap="">No, it is a consequence of using 'incomplete recovery' clause of RECOVER command, i.e. any of UNTIL and/or USING BACKUP CONTROLFILE. When you do so, a mark is placed in the controlfile used during RECOVER command execution, which means - "Insist on opening database with RESETLOGS with this particular controlfile". </pre> </blockquote> <pre wrap=""><!----> ....I accept your description, but it's strange as, during recovery, I used "recover database using backup controlfile until cancel;"
