Re: Not to use SCAN in 11gR2 RAC
Date: Sat, 17 Oct 2009 11:53:12 -0700 (PDT)
I'm not 100% sure that "Official Statements" = best practices (in fact I've seen metalink notes that make me feel pretty secure that Oracle positions on certain things is not quite in alignment with the real world). Why must be have an official statement on everything? Can't we decide a best practice for ourselves? This sounds so much like the small government/big government debate we have going on in the states. Do we need Larry to tell us every little detail on how to run our databases in the form of an "Official statement"?
It sounds to me like you have gotten quite a bit of feedback. As I read that feedback I'm lead to believe that it would be best to use SCAN. Based on the responses I see, and some experience it would seem that the best practice would be to use SCAN. Additionally when Oracle makes it hard to NOT use something, it seems to me that there is an underlying meaning there which is, basically, use the blasted thing. Sure, I can kill CKPT and the database will still run (for example) but why bother?
So the question is, why the resistance? Why are you dead-set on NOT using SCAN and looking for the slightest justification in the face of what seems to be an overwhelming message to use it?
Robert G. Freeman
Ask me about on-site Oracle Training! RMAN, DBA, Tuning, you name it! Author:
Oracle Database 11g RMAN Backup and Recovery (Oracle Press) - ON IT'S WAY SOON! OCP: Oracle Database 11g Administrator Certified Professional Study Guide (Sybex) Oracle Database 11g New Features (Oracle Press) Oracle Database 10g New Features (Oracle Press) Other various titles
- Original Message ---- From: Yong Huang <yong321_at_yahoo.com> To: oracle-l_at_freelists.org Sent: Sat, October 17, 2009 7:39:58 AM Subject: Re: Not to use SCAN in 11gR2 RAC
Thanks for everybody's response. I tried to find official statement on Metalink and database documentation. Didn't find any. The closest is "Oracle recommends that you configure clients to use SCANs, but it is not required that you use SCANs." If anybody can find reference, please post.
I tried to use "srvctl stop listener -l LISTENER_SCAN1" to stop the SCAN listener and got an error that explicitly says SCAN listener can't be stopped. "kill <SCAN tnslsnr IP>" worked. After that, I can continue to use the connection string that resolves to the VIP's. I can't tell if there's any harm done. But I'll keep SCAN when I install for production, for reasons of future updates or patching, and because I didn't find an official statement I can remove it. I opened an SR on this.
On Oct 16, 2009, at 11:59 AM, Yong Huang <yong321_at_yahoo.com> wrote:
> We installed 22.214.171.124 RAC on a RHEL 5.3 two-node cluster
> successfully. It looks like SCAN (Single Client Access Name) must
> be assigned during installation.