From: "Howard J. Rogers" Newsgroups: comp.databases.oracle.server References: Subject: Re: standby database and nolog unrecoverable issue Lines: 45 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 X-Original-NNTP-Posting-Host: 210.50.52.166 Message-ID: <3af48c8e$1@news.iprimus.com.au> X-Original-Trace: 6 May 2001 09:28:14 +1000, 210.50.52.166 Date: Sun, 6 May 2001 09:28:12 +1000 NNTP-Posting-Host: 203.134.64.67 X-Trace: news0.optus.net.au 989105294 203.134.64.67 (Sun, 06 May 2001 09:28:14 EST) NNTP-Posting-Date: Sun, 06 May 2001 09:28:14 EST Organization: CWO Customer - reports relating to abuse should be sent to abuse@cwo.net.au "ff" wrote in message news:WoWI6.17047$Aj1.6639370@typhoon2.ba-dsg.net... > After we failover to our standby database, I found some data are not > available, oracle complains that datafile are corrupted because nologging > was used for some index created at production database. > > Does it mean that when you are using standby database, you can not use > nologging for DDL at all? Otherwise, how can you send the redo to the > standby site? > No, it means that if you use nologging operations on your primary, you need to think carefully how to propagate those things over to the standby. The standard advice is to take a copy of the datafiles for that tablespace, offline the affected datafiles on the standby and drop the tablespace, transfer the data file copies and current archived logs to the standby and perform recovery at the standby. Having never operated a standby myself, however, I can't vouch for whether the standard advice actually works or not... Because the datafiles are freshly copied across, the unlogged data is in them, and the standby is receiving 100% of what it needs. The non-standard advice might well be that after performing a nologged operation, you should seriously consider re-creating the standby from scratch. If you're doing a lot of those sorts of operations, though, that probably is a little unrealistic. Anyway: the summary version of all this is that nologging operations do pose a significant problem for standby databases, and need to be handled with care. Regards HJR > You input is highly appreciated. > >