Hello Niall,
I observed quite some time ago, that sqlplus connections in combination
with LDAP is getting slower and slower with every release.
The reason for it, is that Oracle does more and more LDAP searches with
every release.
With 10gR2, 1 LDAP search was sufficient, with 11gR2 you have already 2,
and with 12cR1 and higher you got 3. For more information, that a look at:
https://blog.dbi-services.com/oud
-oracle-unified-directory-11-1-2-3-oracle-generates-more-and-more-ldap-
lookups-with-every-release/
Unfortunately, there is no Patch or workaround that I am aware of, to fix
this issue.
Cheers,
William
On Tue, May 29, 2018 at 6:25 PM, Stefan Knecht <knecht.stefan@xxxxxxxxx>
wrote:
Not a windows guy at all, but have you tried tracing it?
https://stackoverflow.com/questions/3847745/systrace-for-windows
Might show you the syscall it's hanging on which would confirm if it's
indeed some sort of lookup that hangs.
Stefan
On Tue, May 29, 2018 at 9:11 PM, Niall Litchfield <
niall.litchfield@xxxxxxxxx> wrote:
Hi
Has anybody else observed initial connectivity with the 12.2
windows client to be significantly slower than in 12.1 (or 11.2). My setup
has ldap and TNS_ADMIN involved.
12.1
ORACLE_HOME: C:\APP\ORACLE\INSTANTCLIENT_12_1
TNS_ADMIN: C:\APP\TNS
12.2
ORACLE_HOME: C:\APP\ORACLE\INSTANTCLIENT_12_2
TNS_ADMIN: C:\APP\TNS
Extract from an admin level sqlnet trace in 12.1
(12180) [29-MAY-2018 14:53:56:171] nnflgetnlpactx: NLPA context
successfuly initialized
(12180) [29-MAY-2018 14:53:56:172] nnflgetnlpactx: exit
(12180) [29-MAY-2018 14:53:56:517] nnfloidinfocache: entry
(12180) [29-MAY-2018 14:53:56:517] nnfloidinfocache: Info: Inserted
value DIRECTORY_SERVER_TYPE=OID at index 0 into NLPA_CACHE
and from 12.2
(9712) [29-MAY-2018 14:47:01:777] nnflgetnlpactx: NLPA context
successfuly initialized
(9712) [29-MAY-2018 14:47:01:777] nnflgetnlpactx: exit
(9712) [29-MAY-2018 14:47:38:195] nnfloidinfocache: entry
(9712) [29-MAY-2018 14:47:38:195] nnfloidinfocache: Info: Inserted value
DIRECTORY_SERVER_TYPE=OID at index 0 into NLPA_CACHE
Note the extra 30 secs between exit from the nnflgetnlpactx function and
entry to the OID related function nnfloidinfocache
I have a ticket with Oracle Support but wondered if anyone else has seen
this.
--
Niall Litchfield
Oracle DBA
http://www.orawin.info
--
//
zztat - The Next-Gen Oracle Performance Monitoring and Reaction Framework!
Visit us at zztat.net | @zztat_oracle | fb.me/zztat | zztat.net/blog/
