DSXchange

gunzip will correctly decompress the file on windows if you have it installed. You are most likely seeing an issue because UNIX file line termination is usually <LF> while windows default is <CR><LF>. If that isn't your problem, perhaps you could explain why you expect different numbers of rows.

That is a conversion that will happen implicitly in PX jobs, but you can also explicitly convert using the DecimalToString() function,.

The "OR" condition is not supported in the key expression. You will need to do both lookups and then do the OR logic in the derivations of the output. Unfortunately conditional lookups don't exist in server jobs. If you are doing lookups onto a relational source you could use user-defined SQL to do ...

Use an external command stage with the command "gunzip {filename} -". That way you don't need to store the expanded file on disk and won't get the disk full condition (at least not from unzipping, you might get it when you write your output data)

welkh - In order to post a question you should a) use search to see if it has been asked before b) if it hasn't been asked or answered, decide which forum to post in c) go to that forum and post a meaningful subject line and include all necessary details in your problem description. You should also ...

The sequential file stage in server jobs will read the whole file, you cannot have it read just a portion. If you really need to limit yourself to a given number of rows per execution and you wish minimize passes through your source file then I would suggest you look into the split command to do thi...

The whole hashed file is loaded into memory before the first row is read.

There is no clean way of breaking out of this. I've been frustrated many a time when I made a metadata error and ended up waiting 10-15 minutes for a "view data" to return the 1st row back to me but needs to read to the end of a large file. This would make a good enhancement request.

I don't know where you can change the value you are looking for. I haven't found it in the ENV variables or documented anywhere. If all warnings to go to the log file it might slow the job speed from "fast" all the way down to "glacial" or even slower, so it would be best to cater for this type of w...

The format of a zipped file is binary. Various algorithms including Lempel-Ziv-Welch are used that take advantage of the fact that different byte or group of byte combinations occur with varying frequencies within a file and that not all 256 possible values in a byte are used throughout the file (hu...

Hello manojbh31 and welcome to DSXChange. In order for anyone here to help on this problem you will need to supply some additional information. I assume that you are using the "Interprocess Stage" in your server job. The number of rows that go into this stage will be the same as the number coming ou...

You need to set up your environment correctly in order to do that. The $PXHOME/bin directory must be part of your PATH environment variable, otherwise specify the full path to that program. You will also need to specify an APT_CONFIG_FILE in your environment and running $DSHOME/dsenv should set up a...

I would use a User Variables Activity stage to create a variable with the full command line for the dsjob command, including using the aforementioned function to generate the date as a parameter to that job. Then use the variable in the subsequent command activity stage.

Do you want the timestamp to be part of the filename or part of the file contents?

That derivation is why you are getting dropped records. You need to handle nulls in that derivation as well as in the one you posted earlier. You need to add a IF IsNull... to that clause.