o.k., so even after running adpreclone, we still get ORA-01194: file 1 needs more recovery to be consistent. I found a document re: using this with Oracle ASM, which we haven't been using. Will we have to configure ASM in Oracle for this to work?
Are you trying to snap your application servers or your database servers? I thought adpreclone was for duplicating an app server instance. I worked with a group in the past that put the DB in hot-backup mode (getting out of my area of expertise here and really have no idea what that involved..but maybe this will lead you in the right direction) and then took the snap (using admsnap) which they then mounted and ran their backups.
EMC Replication Manager has support for snapping Oracle DBs but can't comment on how well it works as I've never had an opportunity to work with it.
So, we are an Oracle eBusiness Suite shop. We are running into this problem on the database tier... but run adpreclone on both database and apps tier. We have also tried putting the db in hot backup mode, same result. I found a white paper:
EMC CLARiiON SnapView and MirrorView for Oracle Database 10g Automatic Storage Management
Best Practices Planning
with the following:
Case 2: Using SnapView/clones for database cloning
While Oracle is open for read/write on the local host, a SnapView/clones consistent set split is performed on the established clones to create a restartable image of the active database that can serve as a repurposed database.
Create point-in-time consistent restartable clone set split
It is important to remember that in order to support restart, all the ASM group LUNs (003-014) have to be included in the list for this command invocation.
So, it looks like we should be able to do this... we're just not getting there as fast as I hoped...
I'm assuming you are cloning both the data, redo, and other Oracle-related files in a consistency group and fracturing the clone consistently across all of these and the error you are getting is when you try to bring up the database - correct? This ensures the control files, redo logs, etc are in a sync with eachother.
Once you are sure this is the case, I would open a tar with Oracle regarding bringing up the environment.
As I mentioned before, Replication Manager may (and someone chime in that knows more about this) automate the whole thing.
First, to clarify, we're doing snapshots not clones (EMC Snapshot process, Oracle clone process) though. Yep, we're getting it all data, redo... but I think you got it... We were in default (persistent) mode. We're going to try again in consistency mode. I hope that'll do it. I'll let you know.Thanks Dan!
O.k., snapped in consistent mode, still getting db consistency errors. Can't get an Oracle clone done... keep getting:
Connected to an idle instance. ORACLE instance started.
Total System Global Area 1069252608 bytes Fixed Size 2121640 bytes Variable Size 431259736 bytes Database Buffers 624951296 bytes Redo Buffers 10919936 bytes
Control file created.
alter database open resetlogs * ERROR at line 1: ORA-01194: file 1 needs more recovery to be consistent ORA-01110: data file 1: '/u62/oracle/trng/data/system01.dbf'
ORA-01194: file 1 needs more recovery to be consistent
You need to run recovery, something like : recover datafile 1;
(and please do your homework on this as my Oracle is quite stale)
I think the confusion here may be that a snapshot is NOT the same as taking a copy of a database cold - so when you want to recover it, you have to run recovery. The redo log is written to before the data file, so when you take the snapshot if there may be something in the redo log that isn't in the datafile yet...
Again, if you have an open support agreement with Oracle I highly suggest that you open a case with them as this is an application-specific issue not really a Clariion issue at this point (in my opinion). Also, as I said before, Replication Manager may be a better way to automate this.
But, we're not trying to start the PROD instance. We're trying to clone to a instance called TRNG. We're running PERL adcfgclone.pl dbTier when we get this error. Thanks for your responsiveness! I hope we can figure this out...
I believe that the Consistency Snap is what you are suppose to use and I believe you are suppose to put the DB into something called Hot Backup? That is suppose to pause the DB to allow the snap to collect all the data on all the LUNs in use at the same time - making the snap consistent across all the LUNs. That's about as much as I know about DB and consistant Snaps - there is a White Paper on Snapview that I believe goes into this topic.
o.k., so we have done a consistent snap with the db in hotbackup mode. we then ran dbtechstack... AND tried to create a new control file... but we couldn't get it created. Aaargghhh! We tried to create a new control file as PROD (thinking if we could get it cloned, we could rename later) and TRNG. We JUST can't get this working!
We got it! Finally... with a little help from my hero Bob Ng. Our trouble was that we were trying to begin the Oracle clone process from the snapshot without ever just starting the database. Because the Snapshot is like a hard crash, we had to start the database as it was, and bring down gracefully. After doing that the Oracle clone process worked swimmingly!
judyo1
8 Posts
0
April 23rd, 2010 05:00
o.k., so even after running adpreclone, we still get ORA-01194: file 1 needs more recovery to be consistent. I found a document re: using this with Oracle ASM, which we haven't been using. Will we have to configure ASM in Oracle for this to work?
DanJost
190 Posts
0
April 23rd, 2010 07:00
Are you trying to snap your application servers or your database servers? I thought adpreclone was for duplicating an app server instance. I worked with a group in the past that put the DB in hot-backup mode (getting out of my area of expertise here and really have no idea what that involved..but maybe this will lead you in the right direction) and then took the snap (using admsnap) which they then mounted and ran their backups.
EMC Replication Manager has support for snapping Oracle DBs but can't comment on how well it works as I've never had an opportunity to work with it.
Hope this helps
Dan
judyo1
8 Posts
0
April 23rd, 2010 11:00
So, we are an Oracle eBusiness Suite shop. We are running into this problem on the database tier... but run adpreclone on both database and apps tier. We have also tried putting the db in hot backup mode, same result. I found a white paper:
EMC CLARiiON SnapView and MirrorView for Oracle Database 10g Automatic Storage Management
Best Practices Planning
with the following:
Case 2: Using SnapView/clones for database cloning
While Oracle is open for read/write on the local host, a SnapView/clones consistent set split is performed on the established clones to create a restartable image of the active database that can serve as a repurposed database.
Create point-in-time consistent restartable clone set split
# naviseccli –h array1
–user {admin_user} -password {admin_password} snapview
–consistentfractureclones -CloneGroupNameCloneId
name1 id1 ... nameN id
It is important to remember that in order to support restart, all the ASM group LUNs (003-014) have to be included in the list for this command invocation.
So, it looks like we should be able to do this... we're just not getting there as fast as I hoped...
DanJost
190 Posts
1
April 23rd, 2010 12:00
I'm assuming you are cloning both the data, redo, and other Oracle-related files in a consistency group and fracturing the clone consistently across all of these and the error you are getting is when you try to bring up the database - correct? This ensures the control files, redo logs, etc are in a sync with eachother.
Once you are sure this is the case, I would open a tar with Oracle regarding bringing up the environment.
As I mentioned before, Replication Manager may (and someone chime in that knows more about this) automate the whole thing.
Dan
judyo1
8 Posts
0
April 23rd, 2010 13:00
First, to clarify, we're doing snapshots not clones (EMC Snapshot process, Oracle clone process) though. Yep, we're getting it all data, redo... but I think you got it... We were in default (persistent) mode. We're going to try again in consistency mode. I hope that'll do it. I'll let you know.Thanks Dan!
judyo1
8 Posts
0
April 26th, 2010 11:00
O.k., snapped in consistent mode, still getting db consistency errors. Can't get an Oracle clone done... keep getting:
Connected to an idle instance.
ORACLE instance started.
Total System Global Area 1069252608 bytes
Fixed Size 2121640 bytes
Variable Size 431259736 bytes
Database Buffers 624951296 bytes
Redo Buffers 10919936 bytes
Control file created.
alter database open resetlogs
*
ERROR at line 1:
ORA-01194: file 1 needs more recovery to be consistent
ORA-01110: data file 1: '/u62/oracle/trng/data/system01.dbf'
DanJost
190 Posts
0
April 26th, 2010 12:00
I think the answer to your question is here:
ORA-01194: file 1 needs more recovery to be consistent
You need to run recovery, something like : recover datafile 1;
(and please do your homework on this as my Oracle is quite stale)
I think the confusion here may be that a snapshot is NOT the same as taking a copy of a database cold - so when you want to recover it, you have to run recovery. The redo log is written to before the data file, so when you take the snapshot if there may be something in the redo log that isn't in the datafile yet...
Again, if you have an open support agreement with Oracle I highly suggest that you open a case with them as this is an application-specific issue not really a Clariion issue at this point (in my opinion). Also, as I said before, Replication Manager may be a better way to automate this.
Good luck!
Dan
judyo1
8 Posts
0
April 26th, 2010 12:00
But, we're not trying to start the PROD instance. We're trying to clone to a instance called TRNG. We're running PERL adcfgclone.pl dbTier when we get this error. Thanks for your responsiveness! I hope we can figure this out...
kelleg
4 Operator
•
4.5K Posts
0
April 26th, 2010 19:00
I believe that the Consistency Snap is what you are suppose to use and I believe you are suppose to put the DB into something called Hot Backup? That is suppose to pause the DB to allow the snap to collect all the data on all the LUNs in use at the same time - making the snap consistent across all the LUNs. That's about as much as I know about DB and consistant Snaps - there is a White Paper on Snapview that I believe goes into this topic.
glen
judyo1
8 Posts
0
April 27th, 2010 11:00
o.k., so we have done a consistent snap with the db in hotbackup mode. we then ran dbtechstack... AND tried to create a new control file... but we couldn't get it created. Aaargghhh! We tried to create a new control file as PROD (thinking if we could get it cloned, we could rename later) and TRNG. We JUST can't get this working!
judyo1
8 Posts
0
May 3rd, 2010 09:00
We got it! Finally... with a little help from my hero Bob Ng. Our trouble was that we were trying to begin the Oracle clone process from the snapshot without ever just starting the database. Because the Snapshot is like a hard crash, we had to start the database as it was, and bring down gracefully. After doing that the Oracle clone process worked swimmingly!