#3091: Last-minute addition of Unsup-DCCP

#

Tue Nov 06 11:42:40 2007

timm@fnal.gov - Ticket created

Subject:	Last-minute addition of Unsup-DCCP
Date:	Tue, 06 Nov 2007 11:42:17 -0600 (CST)
To:	osg-int@OPENSCIENCEGRID.ORG, vdt-support@OPENSCIENCEGRID.ORG, osg-eb@OPENSCIENCEGRID.ORG, osg-facility@OPENSCIENCEGRID.ORG
From:	Steven Timm <timm@fnal.gov>

Download (untitled) / with headers
text/plain 1.7k

The last-minute addition of the Unsup-DCCP package to the
OSG worker node client had the following unexpected effect
Since Unsup-DCCP apparently is not available for the SL3/x86_64
platform according to the pacman package, the whole worker
node client install errors out on a SL3/X86_64 platform.
I say "apparently" because there's nothing on the VDT matrix
of supported platforms that says anything about DCCP, I am just
going with the error messages I got and what I can find from
looking at the pacman package itself.

This forces us to set

VDT_PRETEND_32=TRUE
and install a full 32-bit worker node client on a 64-bit machine
when in fact we should only have to install the one dccp package
as 32.bit.

This was very shoddy to throw a package in on the very day
of the OSG release and not having tested it. It's not
acceptable and we need a fix. You should not be able to break
a whole OSG package like WN-client just because one piece
of software is unavailable as 32-bit is just unacceptable. WE expect
better. In this case, a 64-bit sl3 package for dccp must be provided.

It turns out the install doesn't work either with VDT_PRETEND_32=true
either because it breaks for other reasons. Which means
any SL3+x86_64 people are out of luck if they want to install
a worker node client. Why do we bother to put 2 months of effort
into participating in the ITB if people are just going to break
everything on the day of the release?

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.

#

Tue Nov 06 12:10:23 2007

timm@fnal.gov - Correspondence added

[Reply]

Subject:	Re: [vdt-support #3091] AutoReply: Last-minute addition of Unsup-DCCP
Date:	Tue, 06 Nov 2007 12:08:49 -0600 (CST)
To:	VDT support tickets via RT <vdt-support@OPENSCIENCEGRID.ORG>
From:	Steven Timm <timm@fnal.gov>

Download (untitled) / with headers
text/plain 4.4k

Here's the error message we are getting:

Downloading [Unsup-Dccp-1.8.1.tar.gz] from
[http://vdt.cs.wisc.edu/software/questions/1.8.1]...
Package [/usr/local/osg-wn:OSG:wn-client] not [installed]:
Package
[/usr/local/osg-wn:http://vdt.cs.wisc.edu/vdt_181_cache:Unsup-Dccp] not
[installed]:
None of the following alternatives succeeded:
Your os is [SL-3] but it must satisfy [linux-rhel-4].
Your os is [SL-3] but it must satisfy [linux-fedora-4].
Your os is [SL-3] but it must satisfy [linux-suse-9].
None of the following alternatives succeeded:
system processor [x86_64] is not equal to [i686].
Environment variable [VDT_PRETEND_32] has not been set.
None of the following alternatives succeeded:
system processor [x86_64] is not equal to [i686].
Environment variable [VDT_PRETEND_32] has not been set.
None of the following alternatives succeeded:
system processor [x86_64] is not equal to [i686].
Environment variable [VDT_PRETEND_32] has not been set.
Your os is [SL-3] but it must satisfy [linux-debian-3.1].
Your os is [SL-3] but it must satisfy [linux-rhel-5].
Your os is [SL-3] but it must satisfy [linux-sl-fermi-4.0].
Your os is [SL-3] but it must satisfy [linux-rhel-4].
Your os is [SL-3] but it must satisfy [linux-rhel-5].
Your os is [SL-3] but it must satisfy [linux-sl-fermi-4.0].
No download for your platform
Pacman install failed
[root@gaia osg-wn]#

As noted in the previous e-mail, VDT_PRETEND_32 isn't an option here
because the system in question is missing a lot of the other 32-bit
compatibility libraries that this would take.

Steve Timm

On Tue, 6 Nov 2007, VDT support tickets via RT wrote:

> Greetings from the VDT support system!
>
> This message was generated automatically in response to the creation of a
> ticket regarding:
>
> Last-minute addition of Unsup-DCCP
>
> Your original request is copied below for reference.
>
> There is no need to reply to this message right now -- someone from the VDT
> team will respond to you as soon as possible. If you wish to view your
> support ticket online, visit:
>
> http://vdt.cs.wisc.edu/rt/Ticket/Display.html?user=guest&pass=guest&id=3091
>
> Your ticket has been assigned an ID as follows:
>
> [vdt-support #3091]
>
> Please include the ticket ID in the subject line of all future email about
> this issue. To do so, you may reply to this message.
>
> Thank you for your interest in the VDT.
>
>
>
> -------------------------------------------------------------------------
> The last-minute addition of the Unsup-DCCP package to the
> OSG worker node client had the following unexpected effect
> Since Unsup-DCCP apparently is not available for the SL3/x86_64
> platform according to the pacman package, the whole worker
> node client install errors out on a SL3/X86_64 platform.
> I say "apparently" because there's nothing on the VDT matrix
> of supported platforms that says anything about DCCP, I am just
> going with the error messages I got and what I can find from
> looking at the pacman package itself.
>
> This forces us to set
>
> VDT_PRETEND_32=TRUE
> and install a full 32-bit worker node client on a 64-bit machine
> when in fact we should only have to install the one dccp package
> as 32.bit.
>
> This was very shoddy to throw a package in on the very day
> of the OSG release and not having tested it. It's not
> acceptable and we need a fix. You should not be able to break
> a whole OSG package like WN-client just because one piece
> of software is unavailable as 32-bit is just unacceptable. WE expect
> better. In this case, a 64-bit sl3 package for dccp must be provided.
>
> It turns out the install doesn't work either with VDT_PRETEND_32=true
> either because it breaks for other reasons. Which means
> any SL3+x86_64 people are out of luck if they want to install
> a worker node client. Why do we bother to put 2 months of effort
> into participating in the ITB if people are just going to break
> everything on the day of the release?
>
> 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.

#

Tue Nov 06 12:57:00 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6848
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 226b

Commit comment:
Adding "ALLOW_X86_ON_X86_64 = 1" so we can use the rhas3/x86 binary on rhas3/x86-64.

Changed files:
A vdt/branches/vdt-1.8.1/Unsup-Dccp/defs

To generate a diff:
svn diff -c 6848 file:///p/vdt/workspace/svn

#

Tue Nov 06 13:04:22 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6849
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 257b

Commit comment:
Adding Unsup-Dccp to VDT-Extra-Tests so we can verify that it installs
correctly everywhere.

Changed files:
U vdt/branches/vdt-1.8.1/VDT-Extra-Tests/VDT-Extra-Tests.pacman

To generate a diff:
svn diff -c 6849 file:///p/vdt/workspace/svn

#

Tue Nov 06 13:08:57 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6850
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 315b

Commit comment:
Unsupported IA64 because we don't have a binary for it.
Note: when it's unsupported it means that the package will install
without error, but will not bring along a binary.

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

To generate a diff:
svn diff -c 6850 file:///p/vdt/workspace/svn

#

Tue Nov 06 13:15:25 2007

rana@fnal.gov - Ticket 3092: Ticket created

[Reply]

CC:	osg-int@OPENSCIENCEGRID.ORG, vdt-support@OPENSCIENCEGRID.ORG, osg-eb@OPENSCIENCEGRID.ORG, osg-facility@OPENSCIENCEGRID.ORG
Subject:	Re: Last-minute addition of Unsup-DCCP
Date:	Tue, 06 Nov 2007 11:10:28 -0800 (PST)
To:	Steven Timm <timm@fnal.gov>
From:	Abhishek Singh Rana <rana@fnal.gov>

Download (untitled) / with headers
text/plain 2.8k

Hi Steve,

I am replying in my capacity as an 'OSG common man', with no hat on.

Your concern is very valid, yet we should empathize with the team of
colleagues who would have taken the risk of adding this package at the
last moment. Agreed, it could have been done better, but it is very
likely that there was no alternative at that moment.

Please allow me to digress - Latest version of much-famed Mozilla
browser, Firefox, loads webpages so slowly as if there is no tomorrow.
Latest version of much-coveted Mac OS X, Leopard, what a nice name,
could not even meow and failed to impress any. Latest Microsoft
Windows, Vista, oh, end of story. You and I know, patches are coming.

My stone will not reach the team that works on the 300th floor of an
Ivory tower and releases such famed and coveted software. But then,
why throw it at my neighbor, just because they live around the corner?

At a technical level, I do agree with your email, and hope that this
package is fixed soon.

Best wishes,
Abhishek

-- On Tue, 11/06/07 11:42am-0600, Steven Timm wrote:

>
> The last-minute addition of the Unsup-DCCP package to the
> OSG worker node client had the following unexpected effect
> Since Unsup-DCCP apparently is not available for the SL3/x86_64
> platform according to the pacman package, the whole worker
> node client install errors out on a SL3/X86_64 platform.
> I say "apparently" because there's nothing on the VDT matrix
> of supported platforms that says anything about DCCP, I am just
> going with the error messages I got and what I can find from
> looking at the pacman package itself.
>
> This forces us to set
>
> VDT_PRETEND_32=TRUE
> and install a full 32-bit worker node client on a 64-bit machine
> when in fact we should only have to install the one dccp package
> as 32.bit.
>
> This was very shoddy to throw a package in on the very day
> of the OSG release and not having tested it. It's not
> acceptable and we need a fix. You should not be able to break
> a whole OSG package like WN-client just because one piece
> of software is unavailable as 32-bit is just unacceptable. WE expect better.
> In this case, a 64-bit sl3 package for dccp must be provided.
>
> It turns out the install doesn't work either with VDT_PRETEND_32=true
> either because it breaks for other reasons. Which means
> any SL3+x86_64 people are out of luck if they want to install
> a worker node client. Why do we bother to put 2 months of effort
> into participating in the ITB if people are just going to break
> everything on the day of the release?
>
> 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.
>

#

Tue Nov 06 13:23:17 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6851
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 349b

Commit comment:
Added dccp.t to test Unsup-Dccp.
The only test we are doing is to make sure that dccp doesn't segfault
when run. We can't do fancier tests now because we don't install
dcache.

Changed files:
A vdt/branches/vdt-1.8.1/VDT-Certification-Tests/vdt/tests/tests/dccp.t

To generate a diff:
svn diff -c 6851 file:///p/vdt/workspace/svn

#

Tue Nov 06 13:30:27 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6852
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 214b

Commit comment:
Added dccp.t to the test sequence.

Changed files:
U vdt/branches/vdt-1.8.1/VDT-Certification-Tests/vdt/tests/tests/test-sequence

To generate a diff:
svn diff -c 6852 file:///p/vdt/workspace/svn

#

Tue Nov 06 14:16:42 2007

roy - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6853
To:	vdt-support@cs.wisc.edu
From:	roy@cs.wisc.edu

Download (untitled) / with headers
text/plain 280b

Commit comment:
Updated skip counts by adding one to all of them. This will account
for the new test in dccp.t that will only run as part of
VDT-Extra-Tests, not VDT.

Changed files:
U tests/trunk/vdt-test-email

To generate a diff:
svn diff -c 6853 file:///p/vdt/workspace/svn

#	Tue Nov 06 15:09:56 2007	roy - Ticket 3092: Merged into

#

Thu Nov 15 13:03:04 2007

kronenfe - Comments added

[Reply]

Subject:	[vdt-support #3091] SVN commit, rev 6887
To:	vdt-support@cs.wisc.edu
From:	kronenfe@cs.wisc.edu

Download (untitled) / with headers
text/plain 290b

Commit comment:
Updated skip counts for 1.8.1 by adding one to all of them.
This will account for the new test in dccp.t that will only
run as part of VDT-Extra-Tests, not VDT.

Changed files:
U tests/trunk/vdt-test-email

To generate a diff:
svn diff -c 6887 file:///p/vdt/workspace/svn

#	Fri Jan 04 10:11:42 2008	roy - Priority changed from (no value) to '3'

#	Fri Jan 04 10:11:58 2008	roy - Comments added		[Reply]
	Download (untitled) / with headers text/plain 32b This was all done a while ago.

#	Fri Jan 04 10:11:59 2008	roy - Status changed from 'new' to 'resolved'

#	Fri Jan 04 10:11:59 2008	roy - Given to roy