Hi John,

I will check the changes tomorrow, for right now I have a local fix that
gets me going.

Thank you for such a quick response and turnaround time. I'm really
impressed by how responsive you are.

-Joel

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of K. John Wu
Sent: Thursday, January 28, 2010 12:39 PM
To: FastBit Users
Subject: Re: [FastBit-users] File sharing violation during reorder on
Windows

Hi, Joel,

Thanks for the suggestion.  Your observation on the need to empty the 
file cache is very perceptive.  It is exactly what we need to do.  The 
function that performs reordering already perform a number of 
operations to remove indexes and related files.  I have just added 
more function calls to clean the file cache further.  In my testing, 
it seems to work ok.  The code has been check into the SVN repository. 
  It would be helpful if you can try it out when you get the chance.

The easiest way to get the new code from the SVN repository would be 
to download the nightly snapshot from 
<https://codeforge.lbl.gov/snapshots.php?group_id=44>.  Since the 
snapshot is only built nightly at about 2-3 AM Pacific Time Zone, you 
will have to wait till tomorrow.  If you would like to get your hands 
on the code sooner, please let me know.  I will be happy to send you a 
tar ball separately.

John

On 1/28/2010 9:04 AM, Dolisy, Joel wrote:
> Hi,
>
> I'm evaluating the impact of reordering a partition before indexing
it.
>
> A file sharing violation occurs on Windows in the call to
> ibis::part::reorder().
>
> The file sharing violation occurs because the data file has been
opened
> as a memory mapped file with read-only sharing access in the call to
> ibis::column:actualMinMax and the method ibis::part::reorderValues
> attempts to open that same file with write access using a call to
UnixOpen()
>
> It looks like adding a call to
> ibis::fileManager::instance().flushFile(...) before returning from
> ibis::column::actualMinMax fixes the issue, but I'm not sure that it
> does not impact performance in order code path.
>
> Thanks,
>
> -Joel
>
>
>
> _______________________________________________
> FastBit-users mailing list
> [email protected]
> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to