<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
<!DOCTYPE bugzilla SYSTEM "http://bugs.gentoo.org/bugzilla.dtd">

<bugzilla version="2.22.7"
          urlbase="http://bugs.gentoo.org/"
          maintainer="bugzilla@gentoo.org"
>

    <bug>
          <bug_id>149226</bug_id>
          
          <creation_ts>2006-09-26 13:42 0000</creation_ts>
          <short_desc>sys-cluster/torque-2.1.2-r1 deletes needed dirs on re-emerge</short_desc>
          <delta_ts>2006-09-27 09:49:35 0000</delta_ts>
          
          
          <classification_id>1</classification_id>
          <classification>Unclassified</classification>
          <product>Gentoo Linux</product>
          <component>Ebuilds</component>
          <version>unspecified</version>
          <rep_platform>All</rep_platform>
          <op_sys>Linux</op_sys>
          <bug_status>RESOLVED</bug_status>
          <resolution>FIXED</resolution>
          
          
          
          <priority>P2</priority>
          <bug_severity>normal</bug_severity>
          <target_milestone>---</target_milestone>
          
          
          
          <everconfirmed>1</everconfirmed>
          <reporter>ian@syndicated-productions.com</reporter>
          <assigned_to>hp-cluster@gentoo.org</assigned_to>
          

      

      
          <long_desc isprivate="0">
            <who>ian@syndicated-productions.com</who>
            <bug_when>2006-09-26 13:42:51 0000</bug_when>
            <thetext>When re-emerging (upgrading, rebuilding) torque, dirs that do not have any configuraiton files in them are cleaned up when the previously installed version is unmerged.

Some keepdirs should fix this, I think.  Changing the &apos;install&apos; command to a &apos;keepdir&apos; command (and making the chroot mandatory) in pbs_createdirs seems to fix it on my system, at any rate.</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>jakub@gentoo.org</who>
            <bug_when>2006-09-26 13:48:48 0000</bug_when>
            <thetext>Don&apos;t restrict bugs without any reason. Just leave the checkboxes alone.
</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>dberkholz@gentoo.org</who>
            <bug_when>2006-09-26 14:05:10 0000</bug_when>
            <thetext>Good catch. We can&apos;t just change it, because the function also gets run in pkg_postinst().</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>ian@syndicated-productions.com</who>
            <bug_when>2006-09-26 14:09:07 0000</bug_when>
            <thetext>yeah, plus only certain dirs need it, and it depends on whether we&apos;re USEing &apos;server&apos; or not..  
</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>dberkholz@gentoo.org</who>
            <bug_when>2006-09-26 14:16:13 0000</bug_when>
            <thetext>I added a little fix into 2.1.2-r1 -- should hit rsync in an hour or so.</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>ian@syndicated-productions.com</who>
            <bug_when>2006-09-27 08:47:12 0000</bug_when>
            <thetext>Almost perfect.  all appropriate dirs are kept, however pbs_server for some reason seems to think that the .keep file in server_priv/queues is a queue file and tries to parse it.  This is actually messing up my real queue on server restart oddly enough :(

I&apos;m working on a patch to make pbs_server ignore dot-files, as it seems super-easy and i can&apos;t see this limitation causing any operational issues..  Would this be better than risking server_priv/queue be removed due to no .keep?</thetext>
          </long_desc>
          <long_desc isprivate="0">
            <who>ian@syndicated-productions.com</who>
            <bug_when>2006-09-27 09:49:35 0000</bug_when>
            <thetext>update -- a browse of the source indicates that que_recov reading an empty file should clean itself up properly, so my issue must be elsewhere (possibly the way queues are malloc&apos;d).. 

So if we dont mind the error message on pbs_server start, i guess nothing needs to change.</thetext>
          </long_desc>
      
    </bug>

</bugzilla>