This is quite a specific post.
I’ve been doing a lot of performance testing lately. One of the tests that I’ve been using to extract a little more performance out of JDE is to look at oracle database 12c.
I’ve been testing many permutations and combinations of client and server, but I started to get the error below: (after installing JDE on a new and existing template enterprise server).
7045/-170809600 MAIN_THREAD Mon Jul 31 09:13:08.024551 jdb_drvm.c460
JDB9900436 - Cannot determine database driver name for driver type "O"
7045/-170809600 MAIN_THREAD Mon Jul 31 09:13:08.024587 jdb_omp1.c1928
JDB9900254 - Failed to initialize driver.
7045/-170809600 MAIN_THREAD Mon Jul 31 09:13:08.024603 jtp_cm.c209
JDB9909002 - Could not init connect.
7045/-170809600 MAIN_THREAD Mon Jul 31 09:13:08.024617 jtp_tm.c1140
JDB9909100 - Get connect info failed: Transaction ID =
7045/-170809600 MAIN_THREAD Mon Jul 31 09:13:08.024630 jdb_rq1.c2452
JDB3100013 - Failed to get connectinfo
This is on an older tools release (EnterpriseOne 18.104.22.168) (ptf.txt in $SYSTEM/bin32).
I tried 100 different things involving my environment variables and .profile, .bash_profile. I messed around with a ton of things, but then thought – wait. This is a 22.214.171.124 tools and I actually put this down on a existing enterprise server with oracle database 12c client (32 bit). And, this database did not exist when this tools was released(well it was not supported).
It turns out that my error above is because JDE wants to load certain dll’s from the oracle client dir, and it cannot do this from a 12c client.
To get around this, I just installed a new copy of the oracle client (126.96.36.199) and hooked this up to JD Edwards. As soon as I did this, viola! the enterprise server is working perfectly. Note also that this 188.8.131.52 client is talking to a 12c database, as there is reverse compatibility between client and server.
Another slightly interesting thing here is that all I did was tar up the client dir from another machine and untar it on this one – no installers (because I had no graphical interface for the oracle install and I also could not be bothered fighting the responseFile for the next 3 months). As soon as I sprayed out the dir on the linux machine, it all just worked! Just remember that this is a POC machine, so don’t stress, I will not run a production environment like this – it’s just good to know.
At the end of the day, I used a template AWS vm (that was built for another purpose), I unzipped the enterprise server install dir (e900) and oracle client, updated tnsnames.ora and the machine just WORKS.
Complete enterprise server in less than 2 hours? Don’t mind if I do!