UNIX for Advanced & Expert Users

Hi

My client connection to RAC Oracle Release 10.2.0.4.0 database suddenly stopped working.

ACPT_1 - works but not with OEM.
ACPT_2 - does not work at all.

Getting error ORA-12564: TNS:connection refused

Any ideas very much appreciated
Thanks

ACPT_1 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = xxx.xxx.xxx.xxx)(PORT = 1521))
(CONNECT_DATA =
(SERVER = DEDICATED)
(SID = ACPT)
(INSTANCE_NAME = ACPT1)
)
)

ACPT_2 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = xxx.xxx.xxx.xxx)(PORT = 1521))
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = ACPT)
(INSTANCE_NAME = ACPT1)
)
)

The extract from the client trace file is below
(5316) [000001 28-SEP-2008 09:29:04:771] snsbitts_ts: acquired the bit
(5316) [000001 28-SEP-2008 09:29:04:771] snsbitts_ts: acquired the bit
(5316) [000001 28-SEP-2008 09:29:04:771] nsmfr: 1644 bytes at 0x513b9e8
(5316) [000001 28-SEP-2008 09:29:04:771] nsmfr: 736 bytes at 0x5155cf8
(5316) [000001 28-SEP-2008 09:29:04:771] nscall: connecting...
(5316) [000001 28-SEP-2008 09:29:04:771] nladget: entry
(5316) [000001 28-SEP-2008 09:29:04:771] nladget: exit
(5316) [000001 28-SEP-2008 09:29:04:771] nsmfr: 297 bytes at 0x5149478
(5316) [000001 28-SEP-2008 09:29:04:771] nsmfr: 212 bytes at 0x514fef8
(5316) [000001 28-SEP-2008 09:29:04:771] nladtrm: entry
(5316) [000001 28-SEP-2008 09:29:04:771] nladtrm: exit
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: error from nscall
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: ns main err code: 12564
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: ns (2) err code: 0
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: nt main err code: 0
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: nt (2) err code: 0
(5316) [000001 28-SEP-2008 09:29:04:771] nioqper: nt OS err code: 0
(5316) [000001 28-SEP-2008 09:29:04:771] niomapnserror: entry
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: entry
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: reporting NS-12564 error as ORA-12564
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: exit
(5316) [000001 28-SEP-2008 09:29:04:771] niomapnserror: exit
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: entry
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: NSG error 1153 outside of range 12500 to 12530 - reporting as internal error ORA-12160
(5316) [000001 28-SEP-2008 09:29:04:771] niqme: exit
(5316) [000001 28-SEP-2008 09:29:04:771] niotns: Couldn't connect, returning 12564
(5316) [000001 28-SEP-2008 09:29:04:771] niotns: exit
(5316) [000001 28-SEP-2008 09:29:04:771] nsbrfr: nsbfs at 0x51487a8, data at 0x513eef0.
(5316) [000001 28-SEP-2008 09:29:04:771] nsbrfr: nsbfs at 0x5148748, data at 0x513f710.
(5316) [000001 28-SEP-2008 09:29:04:771] nsbrfr: nsbfs at 0x3def858, data at 0x513dff0.
(5316) [000001 28-SEP-2008 09:29:04:771] nigtrm: Count in the NI global area is now 0
(5316) [000001 28-SEP-2008 09:29:04:771] nrigbd: entry
(5316) [000001 28-SEP-2008 09:29:04:771] nrigbd: exit
(5316) [000001 28-SEP-2008 09:29:04:771] nigtrm: Count in the NL global area is now 0
(5316) [000001 28-SEP-2008 09:29:04:771] nlse_term_audit: entry
(5316) [000001 28-SEP-2008 09:29:04:771] nlse_term_audit: exit

ORA-12564: TNS:connection refused reported when making a connection using 10g connection manager (cman) and more than 1 hop. For example
Sql*plus ---> cman1 ---> cman2 ---> Database
Hop 1 Hop 2
Client PC Node A Node B Server
Hops are also known as chaining ....
If so look at:
https://metalink.oracle.com/CSP/main...OT&id=393941.1

vbe - isn't a "subscription" required for metalink? the link you posted doesn't work - it says "link expired"

Maybe...
So here is what is on metalink:

Code:

Applies to:
Oracle Net Services - Version: 10.1 to 10.2.0.4
This problem can occur on any platform.
ORA-12564: reported for connections via connection manager
Symptoms

ORA-12564: TNS:connection refused reported when making a connection using 10g connection manager (cman) and more than 1 hop. For example
Sql*plus ---> cman1 ---> cman2 ---> Database
               Hop 1       Hop 2
Client PC      Node A      Node B      Server
Hops are also known as chaining
Cause
Bug 4570768 10.2 CMAN NOT ABLE DO CONNECTION OVER TWO HOPS - ORA-12564



Solution

Bug is fix in the 10.2.0.4 Patch Set. Patch 6810189
One off patches are available via Patch 4570768

Thanks for reply. My fault, I didn't specified the OS. It's SUN 137111-01 sun4v sparc SUNW,Sun-Fire-T200. And this patch is for linux. Another point is that the problem appeared suddently it has been working Ok for 2 years. So something has changed we don't know what ..