<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Hi Glen,<br>
Thanks for replying. The solution seems to be working. There were many
jobs stuck with E state and after making the config file with $usecp
statement, all jobs with E state disappeared.<br>
Cheers,<br>
Abhi.<br>
<br>
Glen Beane wrote:
<blockquote
cite="mid:5caae8c60903310605q2f1430ebsa50fc50a0be7a37b@mail.gmail.com"
type="cite">
<pre wrap="">On Tue, Mar 31, 2009 at 2:08 AM, Abhishek Gupta <a class="moz-txt-link-rfc2396E" href="mailto:abhig@princeton.edu"><abhig@princeton.edu></a> wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Hi Halvor,
Some time ago I had the same problem and some people come up with the
problem saying that its because of rcp implementation in PBS which can be
changed, but I don't know how to change it tp cp or scp which can probably
solve the issue. If you have any idea about it, please let me know.
</pre>
</blockquote>
<pre wrap=""><!---->
if you have network mounted home directories, you can specify cp for
returning the stdout/stderr files by using the mom config parameter
$usecp
on a lot of clusters you can get away with something like this: $usecp
*:/home /home, which means any time torque needs to copy the
stdout/stderr files to a home directory on the jobs submit host it
will do a local copy to home instead
If you need to use ssh, checkout this documentation:
<a class="moz-txt-link-freetext" href="http://www.clusterresources.com/wiki/doku.php?id=torque:6.1_scp_rcp_setup">http://www.clusterresources.com/wiki/doku.php?id=torque:6.1_scp_rcp_setup</a>
</pre>
</blockquote>
</body>
</html>