Re: Help ! ORA-00600 : internal error code
John, You haven't indicated what version of Oracle you are running or what application you where running when you got this error message. These kinds of error messages are best resolved by logging a tar with Oracle support. I am no expert at 600 error messages, but I believe this error is usually triggered when trying to allocate space in the PGA and the alloc fails. It looks like the PGA was trying to allocate around 32M of space in the PGA and it failed. This may be cause by having SHMMAX set low at the OS level. If this is the case you could try to increase SHMMAX to equal the physical memory on your box. If you are running a PL/SQL program it could be the cursor table is growing and causing the problem. It appears when the cursor table structure runs out of space it reallocates the space and doubles the amount of memory from the previous allocation, then it frees up the older memory chunk. Each time the cursor structor becomes full it doubles the memory again until it makes a request of more than 32M of memory and it fails with the ora-600 error. It seems that on some older versions of Oracle there is maximium of 32M on 32 bit versions of oracle (oracle imposed) when requesting memory for this cursor structure. You may be able to fix the problem with SHMMAX and if that doesn't work you could try to reduce the number of cursors needed by the code. Help ! What does this mean? ORA-00600: internal error cod, arguments : [733], [33719116], [pga heap], [], [], [], [], [] John -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: John Dunn INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: RAC/OPS changes
Well you asked for this. Don't read this if you are faint of heart and weak of mind. Oracle9i Real Application Clusters is the next evolutionary step up from Oracle Parallel Server, and is the result of more than 6 years of development, 9 patents, and 18 additional patents pending. Oracle9i Real Application Clusters are unique in that they provide: Out-of-the-box, near-linear scaling transparency Compatibility with all applications, with no redesign required Fast growth clusters, the ability to rapidly add nodes and disk Based on Oracle's Cache Fusion architecture, Oracle9i Real Application Clusters provide transparent application scalability by quickly and efficiently sharing frequently accessed data across all the servers in a cluster, resolving all manners of contention between servers in the process. In the Cache Fusion architecture, read requests may be served by any of the memory caches in the cluster database. In cases where data is being updated, coordination between the caches of each server becomes necessary so that both the data being read and the data being updated are consistent and correct. If the query request is served by a remote cache, then the block is transferred across the high speed cluster interconnect from one node's cache to another. This fusing of the caches happens automatically and is transparent to the application. This transparency is the key technology that provides the fast, efficient scaling of Oracle9i Real Application Clusters. I warned ya, Scott [via Oracle-L digest] From: Don Granaman [EMAIL PROTECTED] Date: Wed, 25 Jul 2001 22:29:47 -0500 Subject: Re: (Fwd) Re: Oracle 8i and clustering I don't know exactly the context in which you heard this (NT cluster specific?), Don, Probably just my bad paraphrase of a dim recollection of a fast/careless read of previous comments on the list. Can you give a brief explanation of what will be new/changed in RAC, and what it means? thanks, ep ...but rest assured that OPS is not going to be replaced, except in name. The current party line is that real application clusters is a radical departure from OPS. It simply isn't true. RAC is a very major upgrade/rewrite and renaming of OPS, not a different technology. There seems to be a *LOT* of misinformation floating around on this issue and much of it seems be coming from Oracle's marketing machine. -Don Granaman [certifiable OPS OraSaurus] -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Eric D. Pierce INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
RE: 9i On Linux
Roger, I don't believe that Oracle9i is certified on Redhat at the moment. There had been a lag in certification between Redhat and Oracle in the last few months. This situation has been resolved and certification has been restarted between Redhat an Oracle. I think you will have to wait for Oracle9i to be certified on Redhat. However Oracle9i is certified on SUSE. Scott Failed to install on rh7.1, PIII-450/256M (512M Swap). The installation consumed 460M Swap at max. Roger Liu -Original Message- Sent: Friday, June 15, 2001 4:21 PM To: Multiple recipients of list ORACLE-L Can any of you just confirm that Oracle 9i requires 500M of memory on Linux? Thank, Waleed -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Khedr, Waleed INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Liu, Roger (R.) INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: Re[2]: About parallel server
Dick, I am not sure which platform you where using for OPS, but you have to use raw device's. Unless you are using OpenVMS or Compaq TRU64 v5.1 you have to use raw devices for datafiles, controlfiles and online redo logs. Archive logs have to go to a file system. Now there are some unsupported things you can do to get Oracle datafiles on filesystems in an OPS environment but the key word here is unsupported. Scott Brian, I was looking at parallel server for an application we were developing last year. In 8.1.6 at least the control and on-line redo (not rollback segment) files no longer had to be on raw devices. And I stand corrected, there is one and only one standard unix command that does work, dd. It's just such a pile of alphabet soup. That's why we hired a couple of top notch Unix admins. They handle it. Dick Goulet -- Reply Separator -- Author: Brian MacLean [EMAIL PROTECTED] Date: 5/30/01 3:50 PM Your point 1: Unless things have changed redo and controlfiles must be raw. When I took the OPS course several years ago and worked with OPS we needed the redo/controlfiles to be on raw so that one instance could recover when another instance failed. Your point 5: The Unix command dd will do raw. -Original Message- Sent: Wednesday, May 30, 2001 3:48 PM To: Multiple recipients of list ORACLE-L Fernando, Replies included in your original mail, but in addition: Parallel server is a separately priced option from Oracle and it is pricey. Second you may need specific software from you OS vendor to coordinate the file sharing between the servers, again an additional expense. Dick Goulet Reply Separator Author: Fernando Papa [EMAIL PROTECTED] Date: 5/30/2001 2:07 PM Hi everybody! I have some questions about parallel server. Now we have only one instance stand-alone (no parallel), but we are thinking to switch to parallel server because we have a couple of sparc 3500 and nobody are using it, and we think it's good for increase our processing power. The problem is I didn't work with parallel server and I have a lot of questions about it: 1) Is mandatory to use raw devices for control files, redo logs data files? -- Data files yes, redo and control files can be on cooked file system. 2) How we transfer our cooked data files to raw devices data files? import/export? or exist another better (fast) method? -- To the best of my knowledge your going to have to rebuild the database from scratch so imp/exp is your only option. 3) If I start with only one node, performance will be the same of one single instance (no parallel)? -- Yes and NO, raw devices run a little faster than cooked files since the OS's buffer cache is not in the middle. 4) Somebody know how to work with raw devices under solaris? any link? I try to found someting in metalink but there's no samples... -- Working with raw devices is very different from cooked file systems. If you don't have an experienced Unix admin you could be in serious trouble. 5) What about backup? I can't put tablespaces in backup mode and copy with cp... maybe it's time to use rman? -- Rman can handle the backups, but a file system level backup is different. CP does not work anymore, nor does fbackup, or tar. You'll need specialized software for the purpose. Thanks in advance! -- Fernando O. Papa DBA El Sitio - Infraestructura (54-11) 4339-3854 -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Fernando Papa INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com --
RE: About parallel server
Rachel, OPS or in 9i RAC (Hate name changes confuses everyone) raw devices are a requirement of the OS not Oracle. So whether you are using 7, 8, 8i, 9i you can use filesystems for OPS (RAC) if the OS supports multiple concurrent mounts on the same filesystem from multiple nodes. This is the case in OpenVMS and Compaq TRU64 v5.1 . If you use other OS's then the only way to share files is through raw devices. Now Veritas has their Cluster Files System(CFS) that allows multiple file systems to be mounted concurrently by multiple nodes. There is certification being conducted by Veritas and Oracle to support OPS in this configuration. There is more information about Veritas Cluster softerware at there site http://www.veritas.com/us/aboutus/pressroom/2001/01-05-15 -2.html . There is also some other information on VOS (Veritas, Oracle, Sun) at http://www.vosInitiative.com/ . Scott I have heard rumors that OPS on 9i will allow you to use cooked files. User group meeting next week, with a presentation on 9i new features. I'll ask Rachel From: Adams, Matthew (GEA, 088130) [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] Subject: RE: About parallel server Date: Thu, 31 May 2001 06:06:38 -0800 Dick, What are you using for your source for this information? This does not jibe with other things I have read. If your not going to use a shared raw device for the online redo logs, how can one instance to instance recovery for another instance that fails? R. Matt Adams - GE Appliances - [EMAIL PROTECTED] Meddle not in the affairs of troff, for it is subtle and quick to anger. -- Data files yes, redo and control files can be on cooked file system. _ Get your FREE download of MSN Explorer at http://explorer.msn.com -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Rachel Carmichael INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: About parallel server
Ray, Oracle9i RAC (Real Application Clusters) is or was (depending on your perspective) Oracle Parallel Server. Compaq is working with Oracle to pre-install and ship Oracle9i on there NT boxes (not sure about TRU64 or OpenVMS platforms), so you just turn it on and it works. Oracle9i replication is another animal all together. There are some major changes to way you can replicate in 9i. Scott I have heard from iouga attendees that there was is something called Rack from oracle/compaq on the horizon with 9i which is the next replication solution. Can folks comment on this? What is the architecture and how does this play with oracle? Is this just an OPS on some alphastation? Our compaq sales droid didn't know the answer. Thanks. On Wed, May 30, 2001 at 02:07:10PM -0800, Fernando Papa wrote: Hi everybody! I have some questions about parallel server. Now we have only one instance stand-alone (no parallel), but we are thinking to switch to parallel server because we have a couple of sparc 3500 and nobody are using it, and we think it's good for increase our processing power. The problem is I didn't work with parallel server and I have a lot of questions about it: 1) Is mandatory to use raw devices for control files, redo logs data files? 2) How we transfer our cooked data files to raw devices data files? import/export? or exist another better (fast) method? 3) If I start with only one node, performance will be the same of one single instance (no parallel)? 4) Somebody know how to work with raw devices under solaris? any link? I try to found someting in metalink but there's no samples... 5) What about backup? I can't put tablespaces in backup mode and copy with cp... maybe it's time to use rman? Thanks in advance! -- Fernando O. Papa DBA El Sitio - Infraestructura (54-11) 4339-3854 -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Fernando Papa INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- === Ray Stell [EMAIL PROTECTED] (540) 231-4109 KE4TJC28^D -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Ray Stell INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
RE: About parallel server
Fernando, If your current resources are limited and the app is partitioned, you should have very few problems implementing the App in OPS (famous last words). If HA is a requirement and fast failover times OPS is a good pick. Just make sure the ends justify the means. Some people have good experiences with OPS and others may not have had great experiences. Scott We have a couple of independent applications running over the same instance now (it's better to do it because our resources cant' support two instances over the same machine), we think if we use parallel server, we can make partition of this applications, one for each node or someting like that... And, of course, we are thinking about automatic failover and high availability over parallel server... -- Fernando O. Papa DBA El Sitio - Infraestructura (54-11) 4339-3854 -Mensaje original- De: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]En nombre de Rachel Carmichael Enviado el: miércoles, 30 de mayo de 2001 23:50 Para: Multiple recipients of list ORACLE-L Asunto: Re:About parallel server Even more importantly, if you are planning on implementing parallel server just to increase your processing power then you are going to be in trouble! If you haven't specifically designed your application for parallel server, you can end up DECREASING performance by increasing locking and pings. This is not something you do lightly once an app has been installed into production. Rachel From: [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] Subject: Re:About parallel server Date: Wed, 30 May 2001 14:48:03 -0800 Fernando, Replies included in your original mail, but in addition: Parallel server is a separately priced option from Oracle and it is pricey. Second you may need specific software from you OS vendor to coordinate the file sharing between the servers, again an additional expense. Dick Goulet Reply Separator Author: Fernando Papa [EMAIL PROTECTED] Date: 5/30/2001 2:07 PM Hi everybody! I have some questions about parallel server. Now we have only one instance stand-alone (no parallel), but we are thinking to switch to parallel server because we have a couple of sparc 3500 and nobody are using it, and we think it's good for increase our processing power. The problem is I didn't work with parallel server and I have a lot of questions about it: 1) Is mandatory to use raw devices for control files, redo logs data files? -- Data files yes, redo and control files can be on cooked file system. 2) How we transfer our cooked data files to raw devices data files? import/export? or exist another better (fast) method? -- To the best of my knowledge your going to have to rebuild the database from scratch so imp/exp is your only option. 3) If I start with only one node, performance will be the same of one single instance (no parallel)? -- Yes and NO, raw devices run a little faster than cooked files since the OS's buffer cache is not in the middle. 4) Somebody know how to work with raw devices under solaris? any link? I try to found someting in metalink but there's no samples... -- Working with raw devices is very different from cooked file systems. If you don't have an experienced Unix admin you could be in serious trouble. 5) What about backup? I can't put tablespaces in backup mode and copy with cp... maybe it's time to use rman? -- Rman can handle the backups, but a file system level backup is different. CP does not work anymore, nor does fbackup, or tar. You'll need specialized software for the purpose. Thanks in advance! -- Fernando O. Papa DBA El Sitio - Infraestructura (54-11) 4339-3854 -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Fernando Papa INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE
Re: How do I corrupt a block
Jared, BBED is available on UNIX, you just have to make it with the make command. cd $ORACLE_HOME/rdbms/lib make -f ins_rdbms.mk $ORACLE_HOME/rdbms/lib/bbed The bbed command can allow you to soft corrupt a block and you can use the dd command to hard corrupt the block. The bbed command does have a password and its a secret ;) Scott It apparently is only on NT, and unless you have the password, which is known only to Oracle Support Personnel, you can't use it. http://www.ixora.com.au/q+a/0101/23224038.htm Jared On Thursday 31 May 2001 12:10, K Gopalakrishnan wrote: Hi ! The simple thing is you can edit the datablocks using BBED editor. It is shipped with Oracle and You need a password to use that utility. You can browse and edit the data blocks. $BBED will give the required details. BUT IT IS DANGEROUS --- novicedba [EMAIL PROTECTED] wrote: hi, This may sound funny. I want to know how to corrupt a block. I want to test the different methods of identifying block corruption, but I don't have sample data blocks. Please help me novice = Have a nice day !! Best Regards, K Gopalakrishnan, Bangalore, INDIA. __ Do You Yahoo!? Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/ -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Jared Still INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: About parallel server
Fernando, You will need cluster manager and shared disk drives to build this cluster. You need at least Oracle 8.1.6 or 8.1.7 Solaris 2.6 Sun Cluster Manager 2.2 Some volume manager preferably Veritas 2.2.1+ or 3.0.4+ Solaris 8 (2.8) Sun Cluster Manager 2.2 or 3.0 ( If you use SC 3.0 then you will need to apply a patch for bug #1289644 and you are limited to 2 nodes) Some volume manager preferably Veritas 3.0.4+ Solaris 7 (2.7) is not supported for OPS The Shared disk environment can use D1000 disk arrays using SCSI. The interconnect can use SUN's SCI interface. After all that is setup you will need to get the Enterprise version of Oracle and install the OPS option. The OPS option will not display until the cluster is configured and running and any necessary patches applied. Now to answer you questions 1) Yes raw devices are mandatory unless you are running OpenVMS or Compaq Tru64 version 5.1 2) The UNIX dd command works on transferring datafile to raw devices. I have also found that the cp command works on copying files to raw devices and vica versa when using Veritas as the volume manager. Then you probably will need to re-create the control file to change the datafile locations and change MAXINSTANCES. 3)In theory if you start the instance with PARALLEL_SERVER=false you should not be using PCM locks. 4) You need a volume manager to manage your raw devices. 5) It doesn't matter whether the database is OPS non-OPS raw or cooked, hot backups work the same way. You put a tablespace into hot backup mode and copy the datafiles. You just have to use the dd command instead of cp or use RMAN. Also Veritas allows you to cp raw devices. Also keep in mind if the application doesn't scale in your current environment then it won't scale in OPS. You will just end up with more performance issues. You will also need to buy licenses for all these products as well. Good Luck, Scott Hi everybody! I have some questions about parallel server. Now we have only one instance stand-alone (no parallel), but we are thinking to switch to parallel server because we have a couple of sparc 3500 and nobody are using it, and we think it's good for increase our processing power. The problem is I didn't work with parallel server and I have a lot of questions about it: 1) Is mandatory to use raw devices for control files, redo logs data files? 2) How we transfer our cooked data files to raw devices data files? import/export? or exist another better (fast) method? 3) If I start with only one node, performance will be the same of one single instance (no parallel)? 4) Somebody know how to work with raw devices under solaris? any link? I try to found someting in metalink but there's no samples... 5) What about backup? I can't put tablespaces in backup mode and copy with cp... maybe it's time to use rman? Thanks in advance! -- Fernando O. Papa DBA El Sitio - Infraestructura (54-11) 4339-3854 -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Fernando Papa INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
RE: Oracle 9i Articles - self tuning, launch delayed to
Eric, Sales and marketing have 1 job and that is to sell and market. These marketing things are teasers not necessarily giving detailed information. I have not checked but there may be some Oracle9i stuff on OLN. I think OLN has a trial membership (for 30 days?). However I am not sure of the detail. I do know that 9i still offers the old way of tuning, however you can have Oracle do it for you if you want. The automatic tuning allows Oracle to adjust memory structures dynamically and when oracle adjusts these structures it write that information to the SPFILE. The SPFILE is the oracle managed init.ora. This dynamic tuning can also be temporary instead of persistent. 9i also brings back the ability to monitor the buffer cache. This will help you in determining if you need to add or subtract buffers. As far as the MTTR goes, you have this in 8i as well. Oracle8i has you specify the MTTR information in blocks where oracle9i has you specify it in seconds, then oracle9i will determine how many blocks and how often the blocks need to be written to the redo logs for faster instance recovery. Oracle managed files are nothing special. Oracle already creates the datafile for you. Oracle9i allows you to specify an init.ora parameter to specify where the datafiles get created by default. This also works for log files. This way you say CREATE TABLESPACE test and oracle will create the datafile in the specified location from the init.ora. Oracle managed UNDO are also just rollback segments that are managed by Oracle and not you. The oracle managed UNDO are still segments just like the ones you create. The oracle managed UNDO(AUTO) has features that you can't get with old way of doing UNDO (Manual). The AUTO undo automatically creates UNDO segments based on init.ora parameters that I haven't quite figured out yet. The AUTO undo also allows retention of UNDO for a specified period of time. I haven't tested this but have been told that AUTO undo can steal extents from other rollback segments if that specific rollback segment needs more space (extents) and cannot extend in the current tablespace. Scott dude, The article is not actually providing details, it is just reporting that Oracle has provided details. Who got them, when, how, where is a different story. :) It is weird how the article doesn't cite any specific sources at Oracle. I even looked in www.oracle.com's press releases page (yuk), and so forth, and couldn't see anything. Although there is a Press Portal at www.oracle.com that seems to require registration prior to access. In terms of general 9i product information, there is a lot of stuff (see below), but I couldn't find *anything* at the first or second level of the 9i info that jumped out at me as being focused clearly explaining self tuning details. The data sheet on Oracle9i Manageability (URL below) refers http://www.oracle.com/ip/deploy/database/9i/continuity/index.html?manageas.html to startup/backup issues, and indicates that a persistent INIT.ORA feature is used to bring self tuned parameter settings across shutdowns. Sounds like an incredibly giant pile of cr*p to me. The assumption they are presumably arguing *against* is that self-tuned features would somehow be non-persistent. The circumstances under which a self tuning database would forget what it had done previously is unexplained. Apparently the people that produce the marketing drivel at Oracle either have no clue about this specific topic, or the material was presented to them in an incoherent maner (or both). At any rate it is exceptionally unhelpful to the reader (especially those without a lot of time to waste) to hae to attempt to make sense out of such garbage. Moving right along... In the Resource Management section (same URL): - self-managing rollback segments (some detail) - multi-block size db files for portability - memory management: * SGA self tuning (buffer cache and shared pool) * SGA tuning advisories * transparent management of working memory for SQL execution by self tuning the initialization runtime parameters controlling allocation of private memory - Oracle Managed Files (auto creation of db files) - control downtime ... specify mean time to recover (MTTR) ... in number of seconds ... coupled with dynamic initialization parameters ... improve database availability (which means ??, I don't know) - new capability ... resumable statements ... temporarily suspend ... operation ... process encounters out of space errors ... fix problem ... resume the operation from the point of interruption ... without disrupting normal database operations [new section:] End-to-End Management of Oracle's Internet Infrastructure ... is continuous system availability, reliability, and performance ... important
Re: 9i Real Application Clusters
Rajnedra, Oracle9i RAC is what we now call Oracle Parallel Server. Oracle9i RAC is still BETA, so there may not be any white papers at the moment. I have looked and have not found any white papers on RAC. Scott Dear Friends Does anyone have any knowledge or experience with Oracle 9i Real Application Cluster stuff? If you have I'd like to hear from you about what it is and your insight. I think it is not out yet, some of you may have had the opportunity to find what it is. ... Thanks in advance Raj __ Rajendra JamadagniMIS, ESPN Inc. Rajendra dot Jamadagni at ESPN dot com Any opinion expressed here is personal and doesn't reflect that of ESPN Inc. QOTD: Any clod can have facts, but having an opinion is an art ! * This e-mail message is confidential, intended only for the named recipient(s) above and may contain information that is privileged, attorney work product or exempt from disclosure under applicable law. If you have received this message in error, or are not the named recipient(s), please immediately notify ESPN at (860) 766-2000 and delete this e-mail message from your computer, Thank you. * -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Jamadagni, Rajendra INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: Oracle9i block structures
Unal, What part of the block sturcture? Segment Header, Datablock Header? There are things in the blocks that are no different than they where in 6,7,8 or 8i. However with some of the new features such bitmap freelists, low highwater mark, high high waterwater, and changes to pctfree. There are some modifications to the block header and the segment header. They way rows are stored in the block pretty much remain the same. This is pretty much what I have learned to this point. Scott Hi, Do you know if Oracle9i block structures are modified ? thanks in advance... -- Use itrprof SQL Analyzer. It formats SQL_TRACE/Event10046 traces and gives tuning advises. It's web based, no download, no configuration. Just click http://www.unal-bilisim.com/products/itrprof/itrprof_index.html -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: Danisment Gazi Unal (Unal Bilisim) INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). -- Please see the official ORACLE-L FAQ: http://www.orafaq.com -- Author: INET: [EMAIL PROTECTED] Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051 San Diego, California-- Public Internet access / Mailing Lists To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).
Re: Rollback Segments
Don, I hate to be a stick in the mud but there are some things I think need to clarified. I can think of no practical(?) use except for parallel server. In parallel server an instance will aquire private rollback segments specified in init.ora. If none (or too few) are specified, it will aquire public rollback segments. Public rollback segments are, in my opinion, only useful in an OPS environment in an active/passive configuration. (I prefer private, even then.) This can't happen unless you specify both public and private rollback segments in the database. If you have only private RBS in the database and you don't have ROLLBACK_SEGMENTS set to a value, the instance won't start. You will get the error message unable to acquire non-system RBS. If you only have public RBS then you will acquire the rollback segments based on the parameter TRANSACTIONS/TRANSACTIONS_PER_ROLLBACK_SEGMENT. If you use the IBM RS6000/SP I would not use public rollback segments. The SP has non-shared disks and using public RBS could cause performance problems across the switch. If you use SET TRANSACTION USE ROLLBACK can cause problems if you use public rollback segments. SET TRANSACTION USE ROLLBACK can cause problems with private RBS in OPS but at least with private RBS you know which node is owns the RBS. There is just general I/O issues using public RBS as well. If you use private RBS you can at least spread out the I/O. As far as combing public and private RBS in the same database, I would say don't. Combining public and private can cause problems in OPS. If a person puts the a public RBS in ROLLBACK_SEGMENTS and another instance acquires the RBS before you start the 2nd instance then the 2nd instance won't start. You get an error message that the RBS is in use. So the statement If none (or too few) are specified, it will aquire public rollback segments can only happen if you use both public and private RBS in the database, which I would strongly discourage. I have seen an OLTP OPS system with activity on both nodes and public rollback segments all around - and crippling pinging on rollback. The huge disadvantage is the OPS overhead if two instances obtain two rollback segments where some set of blocks in one is covered by the same PCM lock as some set of blocks in the other. Since rollback segments are so write intensive, overhead can be tremendous. This can't happen. Rollback segments get there own PCM locks. This means 2 rollback segments can't share the same PCM locks. This statement is true if you use Oracle7 but since Oracle8 rollback get their own PCM locks. Oracle8i also eliminates the need to write the undo information back to disk for the other node to generate CR images. All CR images are handled by cache fusion in Oracle8i. It is easy to avoid this altogether. Simply create two distinct tablespaces (for the sake of discussion, call them RBS1 RBS2) with two distinct datafiles for rollback. Create private rollback segments for one instance in RBS1 and the other in RBS2. Assign very few fixed locks to these datafiles. I agree with this regarding the tablespaces. You should do this your rollback tablespaces whether you have OPS or not. However you don't assign PCM to datafiles containing Rollback segments. Rollback segments get assigned PCM locks. The parameter GC_ROLLBACK_LOCKS=1-5=500REACH assigns the PCM locks to the segment not the file. 1-5 are the Undo segment numbers and if you specify a range then you have to specify the EACH keyword. This ensures each rollback segment gets it own PCM lock. If GC_ROLLBACK_LOCKS is not specified in the init.ora then if defaults to 1:N releasable for each RBS. If you set GC_ROLLBACK_LOCKS=0 the you get 1:1 releasable. Real Application Clusters are going to be a new ball game with system managed rollback and default releasable locks. Fixed fixed locks are not really eliminated, but it seems that using them will disable cache fusion entirely. It will be interesting. This is also correct. Releasable is the default in Oracle9i, however it is the default as well in Oracle8 and Oracle8i. You have to use releasable locks in Oracle9i to get write - write cache fusion for pinging. If you use fixed locks in Oracle9i then you get writes back to disk for pinging. Thanks, Scott -Don Granaman OraSaurus Jenner Mike wrote: Isn't the concept of a private RBS only applicable to parallel server setups? - Mike. -Original Message- Sent: 09 May 2001 16:21 To: Multiple recipients of list ORACLE-L Hi all, This brings me to a stupid question then. What is the true difference between a public and a private rollback segment? Because if I just think about it like a normal human being I would think that if a user made a Private rollback segment then only that user could use it, but of course I know that can't be the case. Kev
