UNIX for Dummies Questions & Answers

I have a utility provided by our vendor to dump data from their system. It is

expsysdb -s prod proddata.dmp

"-s" and "prod" are parameters and "proddata.dmp" is the file name that the data is dumped to (this can be any name). Our current system (AIX 4.3) has a file size limit set to 1 gig and as you can imagine I am running up againts this limit. Is there a way that I can have the output compressed at the same time that it is being dumped to a file? The reason I need to do this is we have set up a new system and I need to transfer the data to this system. I have used named pipes when transfering the data between two applications on the same machine, but not sure how to work this when it needs to be transferred to another machine. For example I used this to load the data under one session:
impexp -f test named.pipe

Then loged in another session (same user id) and started to dump the data like this:
expsysdb -s prod named.pipe

If you can give me some ideas that would be great or even point me to some examples. Thank you.

Try your named pipe trick again. But start a gzip (or whatever other compression program you use) to read from it. Then start your dump program writing to it.

Thank you. I tried this:
mknod lawpipe.pipe p
gzip -c lawpipe.pipe > test.gz

I get the following error:
gzip: lawpipe.pipe is not a directory or a regular file - ignored

Then try:
gzip < named.pipe > outputfile

Perfect. Thank you very much. You have made my life much easier.

Hi,
A couple more questions. Can you explain a little more how the command works? For instance, what do the arrows (<named.pipe>) on either side of the named.pipe signify? Also, is it possible to run this in background? Again, thank you for your help.

command < inputfile

just specifys an input file for command. This makes the shell open the file and run the command with the file already opened. Some commands can open their own input files. But as you saw here, some of them perform screwball tests and refuse to open perfectly good files. So I prefer this syntax.

command > outputfile is the same deal but in the other direction. This is more often the only game in town because lots of commands can't open an output file...the shell must do it for them.

command < inputfile > outputfile &
will place the command in the background, but it exit if you close the session.

nohup command < inputfile > outputfile &
will background the command and protect it from hangup signals. So it will continue if you close the session.