I'm not sure exactly how they're being zipped - that code is pretty well
buried in the application, hence this sanity check post to the list
before I dig in. It's C++ code that uses the zlib dev libraries to
create a zipped output stream that's probably written to like any other
C++ stream.
The files are ~100K. UNIX cmd line gzip, gunzip, etc. work fine on the
PVFS2 volume, it's only output from the protein modeling software that's
corrupt. But I'd like to track down the issue so that I know there's not
an issue with the PVFS setup. Most likely it's a bug in the modeling
software stream implementation.
Cheers,
-crispy
On 12/01/2010 09:32 AM, Michael Moore wrote:
I'm not aware of any issues preventing gzip from working correctly.
How are the files being gzipped? I just ran a gzip/gzip -d test on a
small text file on a PVFS filesystem and it was successful. What are the
file sizes you're working with?
Michael
On Wed, Dec 01, 2010 at 08:52:40AM -0500, Chris Poultney wrote:
All-
I'm running some third-party software that has an option to write output
files as plain text or as gzipped text files using zlib. Gzipped files
written to an ext4 partition are fine, but those written to the PVFS
volume are corrupt and unreadable by any of the gzip tools (the error
message is "not in gzip format"). Plain text output files are fine in
all cases.
Before I spent a lot of time digging through the third-party software
source and making some simple test code, has anyone else run into a
similar issue? I didn't find anything in the PVFS2 docs that seemed
relevant to this particular issue.
Thanks,
-crispy
_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
