We are still experiencing these REP126 errors.

Please see [b][color=#0000FF]this previous thread[/color].

The errors occur when we have multiple OI sessions running at the same time on the same server (say 5 OI sessions), creating reports. The reports use different methods, but usually a Select_Into( listStat, "EDT" ) followed by output written to a temporary file, then the temporary file is output to a PDF using OIPI.

This is part of an end-of-day process. The errors don't occur each night, but often enough to be a problem.

Anyone got any thoughts?

Cheers,

Matthew Jones.

Matt,

You're sure no process is calling a DEFINE_DATABASE anywhere? A quick search through the object code indicates that CREATE_INDEX might be an offender.

The Sprezzatura Group

The Sprezzatura Blog

World leaders in all things RevSoft

I don't know enough about OI internals to be sure just what is being called, but this is what is happening:

We have an OI task processor that runs on the server.

Each task manager, which is an OI session, can launch multiple OI engine sessions on the same server to process tasks.

In the specific situation we are having difficulty with, the task manager launches 5 OI sessions to handle the parallel processing of some 10 or so reports.

So these 5 OI sessions are all running on the same server, in the same Windows session, at the same time, producing reports.

The reports use a Select_Into( listStat, "EDT" ) followed by output written to a temporary file, then the temporary file is output to a PDF using OIPI.

So I don't know how a DEFINE_DATABASE can be called from that process.

Since posting the original message, we have had more REP126 accompanied by a corrupt DBT file.

Obviously I have no idea how this is happening.

As an aside, we also experience PDF corruption when we generate PDF's using this method with OIPI. I posted sample code previously that proves this corruption, but Revelation don't seem to believe me.

If the sample code is run as documented, with the number of engines specified, I am confident it will reproduce the problem.

I am also confident I know the solution to that problem - allowing the temporary folder to be nominated so the temporary files are created in separate folders by processes in the same Windows session.

Any help on either front would be much appreciated, as these issues are causing critical issues for our clients.

Thanks.

Matthew Jones.

Well, I can't comment on the PDF corruption, however we'll do what we can on the REP126.

There's a few ways around this, but they all involve some heavy duty debugging.

The easiest way is to wait until the error popups up, then load the debugger from that session. The return stack should give you some information on where the message is called from.

If that's not possible, because the engine is hidden, then hopefully you're not running in SYSPROG. You can then shell FSMSG or MSG and if the error is REP126, put in some logging calls using RETSTACK().

You might also try an engine log or a profile log, since they could give some information on the callers.

Without looking at your source, it's hard to say if something is creating an index dynamically, or creating some archive or temp tables, which are the most likely culprits.

Archive or temp tables might be created with an import or a LH_FIX call, or even something creating a TEMP DICT, like a copy table.

The Sprezzatura Group

The Sprezzatura Blog

World leaders in all things RevSoft

View this thread on the Works forum...