8.10/8.95_D1 New Install - Package Build Hanging

I am running into an odd issue with a new 8.10/8.95_D1 install (SQL2000/Win2003 Enterprise). Trying to build the very first combo full package, and the R9621 UBE is hanging on the Deployment server. Kicked off fine, created all directories on Deployment and Enterprise servers, Busbuild ran through in a couple of hours, no error logs created. Busbuild closed, and now the UBE is just running (24hrs now), with the Deployment server maxed out at 100%. The ClientPkgBuild.log file is growing steadily, up to 750MB now--making it too large to open in any viewer, so no help there. Install went fine, PORTTEST was good, etc. Any thoughts? BuildLog.txt file from Deployment Server is attached.

Try putting the statement "WaitForBusbuild=Y" in the [INSTALL] section of your deployment server JDE.ini.

It has fixed memrory errors in the past, but I have used it for other errors like the one you are talking about.

Warrning, it may cause your build to take a little longer.

Update: I was able to get the log file open. Essentially, it is throwing an error on every single line:

Thu Nov 17 11:43:24 - BLOB data for record number 3414394 is corrupt. Correct the error. Spec Error: JDB operation failed!. Additional Error Code: 16

Clearly something has gotten corrupted, but again--this is a new install. Only changes are to lay down 8.95_D1.

Attached are the meaningful parts of the Package Build Log. Very strange in that it shows the following line before throwing the BLOB errors:

Spec count from Central Objects RDB: 0

Anyone seen this before?

8.10/8.95_C1 on Oracle 10g1 - the same issue.



Business Function BuildPackage at line 11 for Event 15 in Section Software Package Build Header has caused a memory violation.

What is the platform you are on. Did the relink Business function work
correctly. Try launching the package build for the client seperate. What about
Response line any word from them? I would also change the INI for Wait for
Busbuild Y
Reboot and then launch a new package build
Barb Kramarski

You are not using OLEDB by any chance are you..?

Platform is Oracle 10 on Linux.
The problem remains actual for all cases : with WaitForBusBuild=Y or without, client+server or client only full package. After hours of work R9621 stops with memory violation. BusBuild perfectly does the work without any error.

Did you put the latest Planner ESU on the Deployment Server as well?

The data sources for Central Objects were set up as ODBC, as the documentation (and other docs) stated that there were a host of problems with OLEDB. The other sources were originally OLEDB (for about a day, nothing done on the system at all), then removed and reentered as ODBC connections. Serious pain. Your question occurred to me, as well, but again, the CO connections were never OLEDB, so I would be shocked if spec corruption occurred that easily.

Solution

Finally fixed this one, and am posting it in case someone else needs a solution in the future. It actually was related to the OLEDB/ODBC issue, where all data sources needed to be manually removed and recreated as ODBCs. The Central Object database data sources were mistakenly entered with an incorrect source owner within the Planner machine source, which in turn caused the spec build process to return a message that the specs were corrupt, when in fact they simply couldn't be accessed. Corrected the error and got sucessful builds for both client and server.