smbfs does not report all files

Bruce LaZerte mail at fwr.on.ca
Mon Sep 25 02:48:59 GMT 2000


** Reply to message from Urban Widmark <urban at svenskatest.se> on Sun, 24 Sep
2000 20:45:17 +0200 (MET DST)


>> Oh! There's a lot of message pairs like this :
>> 
>> kernel: smb_receive_trans2: copied, parms=4079 of 4079, data=220 of 4079
>> kernel: smb_receive_trans2: copied, parms=4079 of 4079, data=4079 of 4079
>> 
>> Since they didn't say anything like "ERROR", I was ignoring them. Could that
>> "data=220 of 4079" have anything to do with the missing files?
>
>They are not error messages, just informational that it used an unusual
>code path. The server didn't respond with a single message, it sent 2
>messages that the smbfs "sock" code had to merge (note that this is all
>tcp so there really is no reason to send a response as 2 messages).
>
>I have never managed to get this message and it may very well be that
>there are bugs here. Since you only get this on large shares, even likely
>(only when it has to split a message in 2 sends).

The "share" is a whole partition on the OS/2 server. The sub-directories with a
small number files come through fine. It's sub-directories with lot's of files
that are missing some... 

>Do you get any other smb_ messages? You should get something from the
>directory parsing code.

Nothing else in /var/log/messages

>If not I'll try to make a patch for you that adds some debugging for the
>case when the "fast track" doesn't work and some debugging in the
>directory parsing. Any favourite kernel version?

Well, I've now got source trees for 2.2.15-4mdk, 2.2.17 and 2.2.18-pre9, so any
of these would be fine. 

Send away.

Would this debugging code patch smbclient as well as smbfs? Or are they on
totally different source branches?

Bruce

__________________________________________
Bruce LaZerte 
Grandview Lake in Muskoka
Ontario, Canada  				




More information about the samba mailing list