Re: problem when using shell script to checkout huge binary files (>70MB)

[Stewart Brodie]

In message <address@hidden>
          address@hidden ("Veronica Zhu") wrote:

> I use Java Runtime.getRuntime().exec() to execute a shell script. This
> shell script does cvs checkout of project repositories  into user
> directories for web based file browsing.
> 
> It works great for all files EXCEPT huge binary files whose size is
> about 70 Mega Bytes. The directory, which should contain those 
> large files, was checkout as an empty directory. (No errors observed.)
> However, at the same time, other directories with smaller files 
> were checkout normally with expected files.
> 
> It doesn't look like the Java problem, because I can use the
> same Java function to execute a shell script that copy the huge
> file from one folder to another one. Well, cvs checkout
> works OK for the large files when doing so in win-cvs windows.
> The server is Red Hat Linux 7.1 and uses 
> Concurrent Versions System (CVS) 1.11.1p1 (client/server).

You say no errors were observed - how are you checking for this?  Do you have
the output of cvs itself or are you going by the exit status?

We saw mysterious partially failed checkouts when we had problems with our
server running out of memory when processing directories contain 120MB+
binary files (we had no choice in the matter) which resulted in "protocol
error".  The operation is cancelled but a successfully error code is
returned, because of a bug in src/server.c which stops the error being
reported properly.


-- 
Stewart Brodie, Senior Software Engineer
Pace Micro Technology PLC
645 Newmarket Road
Cambridge, CB5 8PB, United Kingdom         WWW: http://www.pacemicro.com/