Home » Server Options » RAC & Failsafe » ORA-01033 on one node but DB is started (11.2.0.2 SE / Oracle Linux 5.5 64b 2.6.32-200.13.1.el5uek)
ORA-01033 on one node but DB is started [message #555321] Wed, 23 May 2012 03:40 Go to next message
sei-lks
Messages: 3
Registered: May 2012
Junior Member
Hello,
we encounter a problem with a 2-nodes RAC with 11gR2 SE :
Database RAC is correctly running but we encounter problems using RAC TNS entry, obtaining the message ORA-01033.
We can access to the DB using a TNS entry using each one of IP addresses but when we use the SCAN address, it fails.
2 weeks ago RAC2 node could connect perfectly with the SCAN address but RAC1 (used as a client) obtained error message almost 30% of tries.
We made a reboot of RAC1 node and now SCAN address works correctly from RAC1 and RAC2 (used as a client) fails systematically with SCAN address but works perfectly with direct access to the database.
On both nodes, SCAN address is perfectly resolved by DNS server, cluster seems correct, listeners seems correct ...
We have opened an SR on Metalink 4 weeks ago but we don't progress.

Any idea ???
Re: ORA-01033 on one node but DB is started [message #555336 is a reply to message #555321] Wed, 23 May 2012 04:55 Go to previous messageGo to next message
John Watson
Messages: 4081
Registered: January 2010
Location: Global Village
Senior Member
Welcome to the forum. Please read our OraFAQ Forum Guide and How to use [code] tags and make your code easier to read

I don't understand your problem: ORA-01033 is nothing to do with the SCAN. You need to give a better description of what ishappening.
Re: ORA-01033 on one node but DB is started [message #555340 is a reply to message #555336] Wed, 23 May 2012 05:20 Go to previous messageGo to next message
sei-lks
Messages: 3
Registered: May 2012
Junior Member
Hi John, thanks for your reply.

Sorry for my poor english, I will try to explain better.

We have 2 nodes RAC1 & RAC2 with :
2 public IP
192.168.100.21 OL-RAC1
192.168.100.22 OL-RAC2
2 VIP
192.168.100.11 OL-RAC1-vip
192.168.100.12 OL-RAC2-vip

And an IP address for RAC resolved by DNS :
nslookup OL-SCAN
Server: 8.8.8.8
Address: 8.8.8.8#53

Non-authoritative answer:
Name: OL-SCAN
Address: 192.168.100.14
Name: OL-SCAN
Address: 192.168.100.15
Name: OL-SCAN
Address: 192.168.100.13


Then we have defined a TNS entry on clients (that what I called "RAC" TNS:
DB =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = OL-SCAN)(PORT = 1521))
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = db.myDomain)
)
)
db.myDomain is the RAC instance name, and db1 and db2 the names of each instance in each node.

I obtain the error message ORA-01033 when I do a sqlplus myUser/myPasswd@DB from user "oracle" on node 2.
If, on the same server, I do a direct connection using ORACLE_SID=db2
sqlplus myUser/myPasswd, it works.

If I do the same on node 1 :
sqlplus myUser/myPasswd@DB works fine (now)
sqlplus myUser/myPasswd works fine (with ORACLE_SID=db1)

If I access my database from another client :
It works perfectly using a TNS pointing directly to db1 on node 1 or db2 on node 2, but if I try to use "RAC" TNS "db" (the same as above), some connections works and others fails ... I imagine it works when OL-SCAN is resolved on node 1 and that it fails when OL-SCAN is resolved on node 2.

I hope my explanations are more comprehensive ...





Re: ORA-01033 on one node but DB is started [message #555348 is a reply to message #555340] Wed, 23 May 2012 05:54 Go to previous messageGo to next message
John Watson
Messages: 4081
Registered: January 2010
Location: Global Village
Senior Member
First, check that no listeners are running from a database home. They must all run from the GI home. If that isn't the problem, my guess is that someone has tried to edit some files by hand. That will break the system.

I would re-create the listening environment:

srvctl stop scan_listener
srvctl stop listener
srvctl stop scan
srvctl remove scan_listener
srvctl remove scan
srvctl remove listener

srvctl add scan
srvctl add scan_listener
srvctl add listener
srvctl start scan
srvctl start scan_listener
srvctl start listener


then delete all the tnsnames.ora files and re-create them, too. You may have to delete the listener.ora files, too.
Hope this helps.
Re: ORA-01033 on one node but DB is started [message #555357 is a reply to message #555348] Wed, 23 May 2012 06:55 Go to previous message
sei-lks
Messages: 3
Registered: May 2012
Junior Member
All listener processes were launched from grid environment.
I will try to re-create listeners but I can't do this immediately because servers ar on production.
Thanks
Previous Topic: RAC Interconnect Transfer rate vs NIC's Bandwidth
Next Topic: difference between crsctl and srvctl
Goto Forum:
  


Current Time: Wed Apr 16 03:20:16 CDT 2014

Total time taken to generate the page: 0.13893 seconds