#3007: bug in gums-host-cron

#

Fri Oct 05 13:35:12 2007

timm@fnal.gov - Ticket created

CC:	osg-int@OPENSCIENCEGRID.ORG
Subject:	bug in gums-host-cron
Date:	Fri, 05 Oct 2007 13:25:30 -0500 (CDT)
To:	vdt-support@OPENSCIENCEGRID.ORG
From:	Steven Timm <timm@fnal.gov>

Download (untitled) / with headers
text/plain 1k

It was previously reported on the osg-int that the following logic
error exists in gums-host-cron:

When it comes time to decide whether to create or update a
grid-mapfile, it compares gsi-authz.conf against a
pre-determined string called PRIMA_CALLOUT_STRING.
The default value of PRIMA_CALLOUT_STRING is the following:

PRIMA_CALLOUT_STRING="libprima_authz_module_gcc32dbg"

Obviously on a 64-bit node, PRIMA has a different callout string:

globus_mapping
/usr/local/vdt-1.8.1/prima/lib/libprima_authz_module_gcc64dbg globus_gridmap_callout

and thus a grid-mapfile is generated when there ought not to be one.

So is there any plan to fix this problem, or do we just have to remember
to re-hack our gums_host_cron every time there is a pacman update?

Steve Timm

--
------------------------------------------------------------------
Steven C. Timm, Ph.D (630) 840-8525
timm@fnal.gov http://home.fnal.gov/~timm/
Fermilab Computing Division, Scientific Computing Facilities,
Grid Facilities Department, FermiGrid Services Group, Assistant Group Leader.

#	Fri Oct 05 14:08:29 2007	roy - Given to roy

#	Fri Oct 05 14:08:29 2007	roy - Cc jpackard@bnl.gov added

#	Fri Oct 05 14:09:05 2007	roy - Priority changed from (no value) to '4'

#	Fri Oct 05 14:10:30 2007	roy - Correspondence added		[Reply]
	Download (untitled) / with headers text/plain 1.1k Steve-- You're absolutely right. I recall extensive discussions about this earlier, and I had thought it was fixed in recent versions of GUMS. Apparently it hasn't been. I've copied Jay Packard on this ticket. Hopefully he can fix it and release a new version of GUMS that we'll put into VDT 1.8.1a. Jay? Thanks, -alain > It was previously reported on the osg-int that the following logic > error exists in gums-host-cron: > > When it comes time to decide whether to create or update a > grid-mapfile, it compares gsi-authz.conf against a > pre-determined string called PRIMA_CALLOUT_STRING. > The default value of PRIMA_CALLOUT_STRING is the following: > > PRIMA_CALLOUT_STRING="libprima_authz_module_gcc32dbg" > > Obviously on a 64-bit node, PRIMA has a different callout string: > > globus_mapping > /usr/local/vdt-1.8.1/prima/lib/libprima_authz_module_gcc64dbg > globus_gridmap_callout > > and thus a grid-mapfile is generated when there ought not to be one. > > So is there any plan to fix this problem, or do we just have to > remember > to re-hack our gums_host_cron every time there is a pacman update? > > Steve Timm > > > >

#	Fri Oct 05 14:10:31 2007	RT_System - Status changed from 'new' to 'open'

#

Fri Oct 05 16:05:06 2007

packardj@rcf.rhic.bnl.gov - Correspondence added

[Reply]

CC:	timm@fnal.gov, jpackard@bnl.gov
Subject:	Re: [vdt-support #3007] bug in gums-host-cron
Date:	Fri, 05 Oct 2007 17:03:51 -0400
To:	vdt-support@OPENSCIENCEGRID.ORG
From:	Jay Packard <packardj@rcf.rhic.bnl.gov>

Download smime.p7s
application/x-pkcs7-signature 3.8k

Download (untitled) / with headers
text/plain 1.3k

John Hover fixed it - it was a trivial fix. I replaced version 1.2.8 on
our repository since this was such a minor fix limited to one file.

Jay

Alain Roy via RT wrote:

> Steve--
>
> You're absolutely right. I recall extensive discussions about this
> earlier, and I had thought it was fixed in recent versions of GUMS.
> Apparently it hasn't been.
>
> I've copied Jay Packard on this ticket. Hopefully he can fix it and
> release a new version of GUMS that we'll put into VDT 1.8.1a.
>
> Jay?
>
> Thanks,
> -alain
>
>

>> It was previously reported on the osg-int that the following logic
>> error exists in gums-host-cron:
>>
>> When it comes time to decide whether to create or update a
>> grid-mapfile, it compares gsi-authz.conf against a
>> pre-determined string called PRIMA_CALLOUT_STRING.
>> The default value of PRIMA_CALLOUT_STRING is the following:
>>
>> PRIMA_CALLOUT_STRING="libprima_authz_module_gcc32dbg"
>>
>> Obviously on a 64-bit node, PRIMA has a different callout string:
>>
>> globus_mapping
>> /usr/local/vdt-1.8.1/prima/lib/libprima_authz_module_gcc64dbg
>> globus_gridmap_callout
>>
>> and thus a grid-mapfile is generated when there ought not to be one.
>>
>> So is there any plan to fix this problem, or do we just have to
>> remember
>> to re-hack our gums_host_cron every time there is a pacman update?
>>
>> Steve Timm
>>
>>
>>
>>
>>

>
>
>

Download (untitled) / with headers
text/html 1.6k

#	Fri Oct 05 16:25:39 2007	roy - Correspondence added		[Reply]
	Download (untitled) / with headers text/plain 393b > John Hover fixed it - it was a trivial fix. I replaced version 1.2.8 on > our repository since this was such a minor fix limited to one file. At the risk of sounding obnoxious: I want to insist on being more principled here. Can this change be GUMS 1.2.9? I really dislike the uncertainty of knowing what we've got when one version string labels multiple versions of the software. -alain

#

Mon Oct 08 13:51:25 2007

cat - Comments added

[Reply]

Subject:	[vdt-support #3007] SVN commit, rev 6751
To:	vdt-support@cs.wisc.edu
From:	cat@cs.wisc.edu

Download (untitled) / with headers
text/plain 204b

Commit comment:
Updated to GUMS 1.2.8a, which should fix the problem noted in the ticket.

Changed files:
U vdt/branches/vdt-1.8.1/defs

To generate a diff:
svn diff -c 6751 file:///p/vdt/workspace/svn

#	Mon Oct 08 15:02:51 2007	cat - Correspondence added		[Reply]
	Download (untitled) / with headers text/plain 119b In VDT 1.8.1a, we have updated to GUMS 1.2.8a which is supposed to fix this problem. VDT 1.8.1a will be released today.

#	Mon Oct 08 15:02:53 2007	cat - Status changed from 'open' to 'resolved'

#	Mon Oct 08 16:37:10 2007	cat - Fixed in 1.8.1a added

#	Mon Oct 08 16:37:14 2007	cat - Stolen from roy