Piotr Wrzosek

"Oracle AI Database 26ai" new features list:
docs.oracle.com/en/database/...
Really it's just 23ai RU10 🤣

Oracle AI Database 26ai replaces Oracle Database 23ai Larry Ellison announced Oracle AI Database 26ai  today in his keynote at Oracle AI World conference in Las Vegas. And I am pretty certain you may want some clarification on what this means exactly.…

Great blog post why and how 23ai is now 26ai - the release numbers explained by @mikedietrichde.com : mikedietrichde.com/2025/10/14/o... @oracleace

Oracle AI Database 26ai replaces Oracle Database 23ai Larry Ellison announced Oracle AI Database 26ai  today in his keynote at Oracle AI World conference in Las Vegas. And I am pretty certain you may want some clarification on what this means exactly.…

Mike put up a blog post explaining all of it late yesterday, just not linked it yet here.
And again we have a release numbering change...

mikedietrichde.com/2025/10/14/o...

Oracle AI Database 26ai Free available for download.

www.oracle.com/database/fre...

I've done my first install.

"Oracle AI Database 26ai Free Release 23.26.0.0.0"

Last Friday I had the honor of taking part in the jubilee concert celebrating the 15th anniversary of the Pruszkowianka Wind Orchestra.
Great band and an unforgettable evening.
www.facebook.com/share/1WHted...

"This section is not visible to customers"
Not really.

Looks like it's not available anymore, my SR regarding this was closed with something like "Internal bug was submitted for investigation, but hey, we got that alternate features "Look-up By Error or Error Code's First Argument" or "Search by Stack Trace" added which replace it.
For real?🤦‍♂️

I just hate OEDA. You can't use it offline despite what the readme says (no, it doesn't save js libraries) and it's so damn bugged it's impossible to get what you need.

AHF 25.7 is marked as obsolete and cannot be downloaded. It's the latest available version. You can download it directly from the AHF Doc ID 2550798.1, but there are no checksums there.
WTH?

Better yet, join next Tuesday's launch webinar!

tanelpoder.com/posts/xtop-t...

MOS AI assistant does that too:
bsky.app/profile/pewu...

I got "There has been a critical error on this website." message only for a long time when visiting your blog.

OK, looks like it's not mentioned directly in changelog.

What version of AU introduced that (and other home_settings options)?

That's even more complicated, it changed in the past over the versions - it worked, then not without some workarounds, as of now it's fixed after my SR.
Full story here:
bsky.app/profile/pewu...

Cannot open SQLcl from SQLDev due to this error, but I can connect successfully from the edit connection window as the password field is not checked for content there.

For me just released SQLDev 25.2.2 finally started to work, I can now connect without getting some error and what's more even secure external password store (oracle wallet) connection type started to work! The only issue is when trying to connect from connections list, it expects to enter password.

#Oracle #Exadata System Software #25ai (25.2) is here! 🚀 Enjoy accelerated AI Vector Search, performance boosts, smarter #Exascale management, and enhanced security (SELinux by default). Thanks to our stellar Engineering & PM teams for making it happen! blogs.oracle.com/exadata/post/exadata252

And turns out the connection to source CDB SID is redirected to the hostname of new CDB, but to the management name not VIP so not working properly too. But anyway why CDB when it's PDB that's being relocated?!?
Either way it screws up EM monitoring and backups.
It's all a mess.

PDB services are running, local listener reports them but scans not, so connections are refused with ORA-12514. LISTENER_NETWORKS is not updated, no chance of connection forwarding. On top of that ORA-00600 [krbbbdr_bad_guid] occurs during backup of tombstone PDB files supposedly due to Bug 36534336

Trying to make PDB relocation in availability max mode working (Doc ID 2771737.1) and it's a mess. PDB relocates successfully, but after opening it source CDB logs following error in alert.log: "ORA-12792: Network transfer operation aborted due to TTC error 3111".

On a downside what hasn't changed is #SQLcl still defaults to thick connection on Windows, even with no client present in path/env, while on Linux it's opposite. Why? 🤷‍♂️
Reported it also in the same SR to no effect...
For reference this thread describes my initial findings: bsky.app/profile/pewu...

Another thing worth to mention is #SQLcl now seems to be not so much dependant on JDBC driver version when using thick connection and works with any 23 client (tested IC 23.5-23.8) as long as it's in the path. I don't find the selected excerpt from release notes to be true, it doesn't work that way.

Finally, after 8 months of waiting the #SQLcl 25.2 release fixes this issue.
It took me great effort to fill the SR, deal with the clueless person handling it for several months, escalate, etc. to reach this goal. To just get back to where it was working in 24.1.
Should be happy but I'm more tired.

They removed it along with /opt/oracle/dbserver/dbms/java/lib/jli
due to switching JDK11 to JDK17 for Exadata software, but the default JDK was 1.8 and should not be touched:
ldd /usr/bin/java
[...]
libjli.so => not found

Just spent several hours of diagnosing why java 1.8 won't start when executed from database on #Exadata 25 only to find out Oracle decided to omit /usr/java/default/lib/amd64/jli from /etc/ld.so.conf.d/dbmmgmt.conf 😠
And I think I know why...

ORA-600/ORA-7445 Troubleshooting tools on MOS are not working for me - at least since yesterday.

Where is the Standalone CVU for Oracle Database 19 on Linux x86-64 ?!?
For HP-UX Itanium CVU version 19.27 April 2025 was already released in Nov-2023 🤣