Fat32 - 1Gb file copy is failing

Volker Lendecke Volker.Lendecke at SerNet.DE
Mon Dec 29 21:40:14 GMT 2008

On Mon, Dec 29, 2008 at 04:03:18PM -0500, Kandukuru_Suresh at emc.com wrote:
> Hi Volker,
>   Thanks for understanding my problem. It is taking too long to fail. I
> found some article with same problem. But there is no solution  for this
> in that.
> http://markmail.org/message/mvcci3luzz7fcuqa#query:samba%20vfat+page:1+m
> id:xkb7u3ycsi52i5de+state:results
> Can you help me  for this?

Probably not with current Samba code I'm afraid. One
long-term goal with Samba is to potentially become fully
asynchronous with regards to posix syscalls. This would not
help with the performance problem, but it might help with
the problem that your clients get a network abort.

To see if this helps, a network trace of the failing
operation would be very helpful. The question is if the
client has started sending smbecho requests while Samba is
working on the ftruncate call. If this is the case, doing
all the async work would be worth it. If the client times
out without having sent smbecho packets, we have to find
other ways.

Can you send that sniff? Please use

tcpdump -i eth0 -n -s 1500 -w /tmp/trace.cap

and send the file /tmp/trace.cap (after bzip -9... :-)).
"eth0" is the interface the server uses.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.samba.org/archive/samba-technical/attachments/20081229/50ef5413/attachment.bin

More information about the samba-technical mailing list