[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: TS Copy Out error



PureBytes Links

Trading Reference Links

While waiting for the next digest to appear I did just what Igor and
Mark Brown
would suggest - copyout in segments.  Each OMZ file was then 20-30 MB.

I suspect what is happening is that the copy-ed data is stored in TEMP
files in the WORK directory (as defined in the OR_Serve.INI file).  ANd
this is uncompressed (the OMZ is compressed), which is why it requires
several times (3 actually) the OMZ file size.  Then when the server
tries 
to compress all those TEMP files into one 300-400MB file it runs out of
space.

I suspect one can change the WORK directory to point to a capacious (and
empty) 
disk for the temp files.  But then one must make a pretty good estimate
of
the storage needed.  Better probably to copy-out in several segments. 
AFTER
making sure you have a hell of a lot of disk space free ;-)

The big problem is you won't know you don't have enough space until
after
most of the work is done, which can take hours. 

Thnaks, guys.

donc

IUhrik@xxxxxxx wrote:
> 
> In a message dated 98-08-23 14:33:37 EDT, you write:
> 
> << I am having a problem Copying_Out the tick data.
>  My tick data file is 470MB.  I am copying out
>  about 60% of the portfolio, enough data to make
>  probably a 250MB file.
> 
>  After two hours of processing, the server says
> 
>  "Not enough disk space to compress d:\....."
> 
>  There is ONE GIGABYTE of free space on this disk ! ! ! !
>  How much !@%#^!@# space does this program need for a 250MB file ? ? ?
>  ....
>  Has anyone else seen and dealt with this errror?
> >>
> 
> What you've experienced is not an error. That's the way it is.
> 
> I know from my experience that to create an .omz file one needs
> free space equal to about twice the size of the ORIGINAL data file
> or of its part that one wishes to copy out.
> 
> All you have to do is to copy out the file in segments. For instance,
> you can do it in 2 parts -- A to M and N to Z, or by date, which I
> happen to prefer. If the split in two parts is not enough, use shorter
> segments. (I use monthly segments).
> 
> Igor Uhrik