powered by Jive Software

Openfire 3.4.5 has been released

We are pleased to announce the release of Openfire 3.4.5! This release includes a brand new database connection pool implementation, which handles database outages in a much smoother manner. It also includes some admin interface robustness improvements and a small memory leak fix. You can view the full change log here.

The SparkWeb client included in the Enterprise edition has been updated too. A wealth of bug fixes and improvements have been made, including support for configuration of MUC rooms, MUC room search support, and autologin improvements. You can use the SparkWeb client from here.

Download Openfire from here.

Download Openfire Enterprise from here.

Download Connection Manager from here


Openfire Team


i just updated from 3.4.4 and I have a small problem: The version shown in the admin panel is still 3.4.4. I never had that problem before and I just followed the steps in the upgrade guide in the documentation. Is this a problem with my installation or is there’s something wrong in the download package?



Well, it shows 3.4.5 for me in Admin Console. Either with tar.gz or exe installation.

Hey Michael,

What do you see in the info.log? If you see there 3.4.5 but 3.4.4 in the admin console then your browser is caching an old page. To force a refresh press Shift-F5. Otherwise, for some reason the lib/openfire.jar file was not replaced with the new one.


– Gato

The info.log shows this line:

"2008.02.09 14:40:42 Openfire 3.4.5 "

But on the entry page of the admin panel version 3.4.4 is shown and there’s the message that a new version is available. I’m using the tar.gz package and I followed the steps in the dokumentation: http://www.igniterealtime.org/builds/openfire/docs/latest/documentation/upgrade- guide.html

I updated several times before without any problems.



Do you have enterprise? If so have you also upgraded the enterprise plugin to 3.4.5?

I can’t imagine why you’d be seeing 3.4.4 though. =/ Unless the old one is still running somewhere? Make sure there’s no lingering java processes from the old version. (or restart the server entirely if that’s easier)

No, I’m running the GPL version. I already restarted the server but the problem still exists. I deleted the entire folder of the former version it’s really weired.

with openfire 3.4.5 GPL ver. i have a problem connecting (or creating a fresh new) to Oracle DB. with openfire 3.4.4 it’s OK.

Database and Version:


Oracle Database 10g Enterprise Edition Release - Production

With the Partitioning, OLAP and Data Mining options

JDBC Driver:

Oracle JDBC driver

JDBC Driver Version:

va Version:

1.6.0_04 Sun Microsystems Inc. – Java HotSpot™ Server VM



OS / Hardware:

Linux / i386

error.log is attached.

thanks for help.

I am have the exact same issue with Oracle on the mac, tried several Oracle JDBC driver versions.

Java HotSpot™ Client VM (build 1.5.0_13-121, mixed mode, sharing)

Not attaching a log, as it matches azagzin’s log file almost verbatim.


putting openfire in debug nets:

2008.02.11 10:03:58 Proxool statistics legend: “s - r (a/t/o)” > s=served, r=refused (only shown if non-zero), a=active, t=total, o=offline (being tested)

2008.02.11 10:03:58 000000 (01/01/00) - Connection #1 created on demand = ACTIVE

2008.02.11 10:03:58 A connection failed the validation test with error: java.sql.SQLException: ORA-00923: FROM keyword not found where expected

Seems that there is an issue with whatever SQL is being used to TEST the connection.

You’re kidding me, ORACLE can’t do a SELECT 1?

I swear…

Ok, for now, you can get around this by editing openfire.xml … look for:


It baffles me that Oracle won’t allow that…

that did it, i ended up using “select * from dual” for testSQL.

thanks for the help.


From dual? What’s that?

Ah, a dummy table. MySQL seems to have it too. I wonder if all of the others have it?

actually, one could do a select 1 from dual as well.

DUAL is a table which is created by oracle along with the data dictionary. It consists of exactly one column whose name is dummy and one record. The value of that record is X. The owner of dual is SYS but dual can be accessed by every user. As dual contains exactly one row (unless someone fiddled with it), it is guaranteed to return exactly one row in select statements. Therefor, dual is the prefered table to select a pseudo column (such as sysdate

select sysdate from dual).

above paraphrased from:


Unfortunately, it appears it is also not standard. =/ MySQL has it, SQL Server, Sybase, and some others don’t. Arg.

I encountered the same problem with Oracle and was about to post my findings here - shame I didn’t wait till tomorrow to perform the upgrade!

I added <testSQL>SELECT 1 FROM DUAL</testSQL> to the defaultProvider configuration.

I don’t suppose I could ask if we have folk using and whether you had success with 3.4.5:

  1. SQL Server (Microsoft’s server)

  2. Sybase

  3. Postgres

  4. IBM’s DB2

Just for my own comfort, I’d like to hear someone give a thumbs up to each of these. From the research I just did, the rest -should- be fine. (IE, Oracle is the only one that doesn’t like a SELECT without a FROM)

jadestorm wrote:

You’re kidding me, ORACLE can’t do a SELECT 1?

I swear…

It baffles me that Oracle won’t allow that…

How did this not get caught by Jive QA before the release?

Good question. All I can really say about it at this point is, we’ve identified a gap in our QA procedures that needs to be addressed.