<br><br><div class="gmail_quote">On Tue, Feb 28, 2012 at 7:01 PM, Wolfgang Bangerth <span dir="ltr"><<a href="mailto:bangerth@math.tamu.edu">bangerth@math.tamu.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">On 02/28/2012 06:08 PM, Thomas Geenen wrote:<br>
> yep that looks like it<br>
> but rather system ("mv tmp_filename filename &");<br>
<br>
</div>I'm already doing this on a background thread, so the '&' is somewhat<br>
implied :-) Your solution is dangerous since if the program terminates,<br>
it will also kill the descendent move job you start here. On a<br>
background job I have the option of at least waiting for the thread to<br>
finish in the destructor before the program terminates.<br></blockquote><div><br></div><div>ah perfect </div><div>i guess that is one of the benefits of doing a proper exception handling.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im"><br>
<br>
> beeing able to set the /scratch dir from the input(prm) file would be<br>
> nice as well.<br>
<br>
</div>Why would you care to set it to something else than the system tmp<br>
directory?<br></blockquote><div><br></div><div>thats just what the sys admin choose i guess to call the local raid partition</div><div>in our case its /local but on other systems its /scratch or /scratch_local etc</div><div>
<br></div><div><div>Filesystem 1K-blocks Used Available Use% Mounted on</div><div>/dev/sda1 19685912 3402564 15283348 19% /</div><div>none 49595884 204 49595680 1% /dev/shm</div>
<div>/dev/md0 882695836 877288 836980192 1% /local</div><div>/dev/sda3 1968400 35768 1832640 2% /tmp</div><div>/dev/sda2 1968400 760376 1108032 41% /var</div></div>
<div> </div><div>in this case /tmp is not the right choice.</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im"><br>
<br>
> an added benefit of doing it like this is that you do not have to wait<br>
> till the end of the run to inspect your results<br>
> for curious people like me thats a huge advantage.....<br>
<br>
</div>Yes. Plus I think it's going to be a problem if many processors all at<br>
the same time start to move *all* of their big files -- at least it's<br>
going to take a while.<br>
<br>
OK, so I think we have two solutions:<br>
- Timo's MPI I/O solution for big machines with fast interconnects<br>
- the background writer job above. I've just implemented this -- take a<br>
look. I do have a problem here, though: all functions like tmpname,<br>
tempnam, mktemp etc are marked as "do not use this function" and<br>
produce a linker warning. The only one one should use is mkstemp<br>
but this function creates a file handle that points to a file that<br>
is destroyed upon closing the file handle. How am I supposed to<br>
copy this file around then if it gets deleted right away?<br></blockquote><div><br></div><div>making the user responsible to set the correct output dir for local storage should solve this right?</div><div><br></div><div>
what i did originally was setting the global output dir (slow home dir, NFS)</div><div>and setting the local output dir. </div><div>in case the local output dir is not set or does not exist etc default to the slow global one</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Cheers<br>
<span class="HOEnZb"><font color="#888888"> W.<br>
</font></span><div class="HOEnZb"><div class="h5"><br>
------------------------------------------------------------------------<br>
Wolfgang Bangerth email: <a href="mailto:bangerth@math.tamu.edu">bangerth@math.tamu.edu</a><br>
www: <a href="http://www.math.tamu.edu/~bangerth/" target="_blank">http://www.math.tamu.edu/~bangerth/</a><br>
<br>
_______________________________________________<br>
Aspect-devel mailing list<br>
<a href="mailto:Aspect-devel@geodynamics.org">Aspect-devel@geodynamics.org</a><br>
<a href="http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel" target="_blank">http://geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel</a><br>
</div></div></blockquote></div><br>