DSXchange

There are 3 performance considerations I can think of for SQL Loader Vs. OCI Load. 1. SQL*Loader is naturally a little faster than OCI load, but the reasons for this are unclear as they use much the same load mechanism. On a remote database, OCI must traverse the network for each buffer of rows, but...

Damn Linux has ruined it for everyone! Just tried the above solution on Linux (SuSE SLES and ps returns heaps more than 255 chars. I tried up to 1000 and I couldn't break it.

Use Ray's solution. Or you could also use the PARFILE option to stick the userid/password in a file.

Most Oracle DBAs know about this one. There is a solution that works with all flavours of Unix I know, and exploits the fact that PS shows only the first 255 chars of the command. Some versions of Oracle actually ship with it included I think (at least they do with sqlplus and sqlforms) - but my cur...

Persist with OCI Load. It works - I use it for all my high volume loads. OK, so you own the table. Clearly the userid, password and database name are correct because you are getting into the database in order to get the 942 error. What's left? I can think of 3 things: 1 - You have mis-spelled the ta...

Thanks guys. Kim, I found the EtlGetRowCountsFromLog job in EtlStats. I can see how I would go about grabbing params from there. Ray, I looked at RT_STATUSnnnn and didn't think it would help that much, but then I opened the file for a restartable job, and lo-and-behold we have a CHECKPOINT row. Look...

We're using the standard checkpoint functionality to make our job sequences restartable. The only problem is that when you restart the top-level sequence, it asks for the parameters to be re-entered. I want an easy foolproof way to restart any restartable job seq using the same parameters as when it...

And in fact you are no worst then you've been so far, as already said, if the 5 Premium posters moved on and stopped posting. Not quite. If the Top5 moved on, there would be (way!) more posts with 0 responses. This is one of the first things that non-gurus look for - a question Ray hasn't gotton to...

I know how I'd go about it, just not 100% on the syntax. Try this: 1. Create a routine ClearFileRoutine with 2 lines: EXECUTE "CLEAR.FILE " : Arg1 Ans = 1 2. Create a Stage Variable ClearFile in your first transformer with the derivation: If IsNull(ClearFile) Then ClearFileRoutine("Fi...

Guru advice costs, fair enough. I've received at least $50 assistance from the top5 in past. Non-guru advice is free. Also fair enough. Sometimes its good, sometimes not so good. Often its too late. But here's the problem: I occasionally connect to kill some time and answer some non-guru questions. ...

Apologies for reopening an old thread, but this might be useful for future searches. This happended to me as well (Linux - 7.5.1A). Trying to run inastall.sh for the purpose of loading some plugins, I got the same error. Turns out that the file system on which DSEngine resides (/opt in my case) was ...

If you are using the stage called "Oracle OCI Load", it doesn't use SQL*Loader in Automatic Mode, so there is no Control file. OCI (Oracle Call Interface) provides a programatic direct-path interface that allows DS to link the load code directly into the plugin. I have another stage called...

I have seen several posts in this forum advising that you can use the UV stage rather than the Hashed File Stage. Since a Hashed File is really just a UniVerse table in disguise, you can use the UV stage to lookup the file using SQL. Note that you may need to index the table/file in order to get acc...

Bad news on the 7.5.1A front Craig. Changed the default to 150 in Director(Tools/Options), bounced server, launched a job: Not only did it stay Unlimited, the default number (should you choose to accept it) was still 50. The mystery continues. My money is on "hard-coded", with a slim chanc...

...how do you run a unix script from Oracle Taking you literally, to run a Unix script from Oracle, see http://www.orafaq.com/forum/mv/msg/55174/147495/78693/#msg_147495 In short, it's easy on Oracle 10g using DBMS_SCHEDULER package, trickier otherwise because you have to write you own launcher in ...

Excellent! That first solution is brilliant. All I need to do is swap my files for IPCs - child's play. It's running at a writeback rate of 600-700 rows/s, which is pretty creditable given the number of IPC's and the miserable config of my dev box. It should do better on a box with more procs. Am I ...