Different Client Access and locking problems

[Luke Kenneth Casson Leighton]

On Wed, 19 Nov 1997 Nikki.Harris at unilever.com wrote:

>      Dear SAMBA Team,
>      
>      I have installed SAMBA on one of our HP-UX boxes running HP-UX 10.01 
>      and locking is working beautifully between the NT clients.  I do 
>      however have a problem with different client access locking problems.  
>      I'll try and explain!
>      
>      I currently have lots of Window for Workgroup 3.11 clients running a 
>      product called PCTCP which allows them to connect to a NFS HP-UX file 
>      system,via IP Stack, where all the Users home and shared data is 
>      stored.
>      
>      I have the challenge (?) of moving all our User PC's to NT whilst 
>      leaving the users data stored on the HP-UX box.  SAMBA has so far 
>      proved to be the best solution for allowing NT clients to access the 
>      data on a HP-UX box ! Wonderful, brilliant!!!!  
>      
>      There is only one problem during the migration phase I will have both 
>      PCTCP clients and NT clients accessing the same data on the HP-UX box 
>      and what I cannot get to work is the locking between the 2 client 
>      access modes , PCTCP via IP/NFS and SMB via SAMBA. 
>      
>      Have you heard of this problem before or have you any clues as to 
>      configuration that may help?
>      
>      Both products, PCTCP and SAMBA, in the documentation talk about using 
>      the rpc.lockd process, but write to different log file - any ideas 
>      would be gratefully received and I am sure that we could arrange Pizza 
>      delivery to Australia from the rainy shores of England!

hi nikki,

pizza?  from england to australia, and then back again?  damn.  it's going to
have its own extra toppings by the time it gets to cambridge... 

right.

we replaced PC-NFS 5.0 under DOS/WfWg 3.11 at my former place of work.  
we had exactly the same problem: file locking issues between the NFS 
daemon and the samba daemon.

files shared between myself (using NT 3.51 accessing samba) and my colleague
(using, ha-ha, WfWg and PCNFS), were inconsistent.  as we were using
SourceSafe (not the visual version), we had to switch off "native locking"
and use the built-in locking mechanism in SourceSafe.  this was sufficient. 


regarding the documentation of the use of rpc.lockd, the documentation states
that samba uses the unix fcntl() call, which on _some_ unixen uses rpc.lockd,
and furthermore that implementations of rpc.lockd can be particularly
bug-ridden...

the documentation also mentions that you might want to try "strick 
locking = yes", which may have a significant read/write performance hit.

regards,

luke


<a href="mailto:lkcl at switchboard.net"  > Luke Kenneth Casson Leighton  </a>
<a href="http://mailhost.cb1.com/~lkcl"> Samba Consultancy and Support </a>