Re: Clarifications: CVS Import Bug - Please Respond

[Derek Price]

Larry Jones wrote:

>Derek Price writes [quoting me]:
>  
>
>>>Since directory
>>>creation is supposed to be atomic, even under NFS, I would argue that
>>>any NFS implementation that returns based on a cache rather than waiting
>>>for the server response is just plain broken.
>>>      
>>>
>>Possibly, but as far as I know it is a common problem.  Don't know why.
>>    
>>
>
>"Common" as in "occurs with lots of different implementations" or
>"common" as in "occurs with lots of different machines, all of which are
>running some variant of Linux"?  If it were the former, I'd have
>expected to have run into it long ago.
>  
>

Common, as in I seem to recall finding lots of references on the web
when I had to solve this problem the first time.  A quick Google search
for "nfs directory caching"
<http://www.google.com/search?q=nfs+directory+caching> brings up high
scoring references to turning off a similar cache with a Solaris NFS
client, a WOE32 NFS client, and an unspecified client from anl.gov. 
Linux shows up around search result #8 or #9.

Cheers,

Derek

-- 
Derek R. Price
CVS Solutions Architect
Ximbiot <http://ximbiot.com>
v: +1 717.579.6168
f: +1 717.234.3125
<mailto:derek@ximbiot.com>