Your JAVA_HOME environment variable in your .bashrc
was set to a PPC version of Java. Build.eclipse.org is now an
x86_64 machine.
I've fixed this for you.
Denis
On 09/23/2010 02:15 PM, Trip Gilman wrote:
It seems to pause directly after or
at the end of the buckminster materialization phase. Also,
I’m getting this error on build.eclipse.org when trying to run
an ant script:
tegilman@build:~>
/opt/public/common/apache-ant-1.7.1/bin/ant -f ~/publish.xml
/opt/public/common/apache-ant-1.7.1/bin/ant: line 329:
/opt/public/common/ibm-java-jdk-ppc-60/jre/bin/java: cannot
execute binary file
/opt/public/common/apache-ant-1.7.1/bin/ant: line 329:
/opt/public/common/ibm-java-jdk-ppc-60/jre/bin/java: Success
If I try to invoke java directly everything seems to be
configured properly.
Trip
On 9/23/10 1:07 PM, "Denis Roy" <denis.roy@xxxxxxxxxxx>
wrote:
That's not normal. It would be
good to know which part of the build is taking longer.
Denis
On 09/23/2010
02:03 PM, Trip Gilman wrote:
FW: [cross-project-issues-dev]
Hudson users: Please attach your builds to slaves Meant to
send this to the whole list. It appears that the build
did finally finish, though the overall build time on slave
2 is about 2X longer than on the master. If this is
normal I’ll just update our docs with the expected time
and continue on.
Trip
------ Forwarded Message
From: Terran Gilman <trip@xxxxxxxxxxxxxxx>
Date: Thu, 23 Sep 2010 11:36:35 -0500
To: David Carver <d_a_carver@xxxxxxxxx>
Conversation: [cross-project-issues-dev] Hudson
users: Please attach your builds to slaves
Subject: Re: [cross-project-issues-dev] Hudson
users: Please attach your builds to slaves
I’ll do that. Also, it appears that the
buckminster-voicetools-nightly job continues to hang at
the materialization stage each time. Even with the debug
setting there are no errors listed, just silence at some
point. Maybe I’m doing something wrong with my
configuration now that I’m on slave 2?
Trip
On 9/23/10 11:16 AM, "David Carver" <d_a_carver@xxxxxxxxx>
wrote:
Trip, best thing to do in
this case is open a bug and ask for it to be configured
appropriately for Buckminster.
Dave
On 09/23/2010 08:38 AM, Trip Gilman wrote:
Re:
[cross-project-issues-dev] Hudson users: Please attach
your builds to slaves We can actually get rid of my
job dependencies completely if the standard platform
packages (rcp/classic/etc) were made available through
the buckminster hudson plugin target platform
mechanism. What we are doing right now is grabbing a
subset of this platform and creating a faux platform
for our other jobs to build against. If I could just
select the proper target (version/package) directly, I
could run my real build jobs as floating on any
available node. Thoughts?
Trip
On 9/23/10 10:20 AM, "David Carver" <d_a_carver@xxxxxxxxx>
wrote:
What Hudson
will do, is use the first available Node for the
build. Subsequent builds will always try to use the
same Node if it is available, if not, it will then
go to one of the other Slave machines. You can
also select to Tie the build to only run on Slave
machines (i.e. Nodes with labels like Linux,
Windows, etc). Then it will try and use one of
those nodes for a build. If your build was tied to
master before, uncheck that option, or choose one of
the other slave machines. It'll help all the other
projects as well.
Dave
On 09/23/2010 07:36 AM, Denis Roy wrote:
Let me try
that again:
The "?" icon next to the option says:
Sometimes
a project can only be successfully built on a
particular slave (or master). If so, this option
forces Hudson to always build this project on a
specific computer. If there is a group of machines
that the job can be built on, you can specify that
label as the node to tie on, which will cause
Hudson to build the project on any of the machines
with that label.
Otherwise, uncheck the box so that Hudson can
schedule builds on available nodes, which results
in faster turn-around time.
This option is also useful when you'd like to
make sure that a project can be built on a
particular node.
So my guess is "yes" your best bet is to leave
it unchecked.
Denis
On 09/23/2010 10:19 AM, Stéphane Bouchet wrote:
Hi,
what if we just untick the "Tie this project
to a node" option ? will hudson choose
automatically the best node ?
Le 23/09/2010 16:04, Denis Roy a écrit :
Folks,
Please configure your jobs to use the Hudson Slave
instances. We've
allocated more memory and CPU resources to the slave than
to the master,
since we were recommended to use the master primarily as a
Hudson
control device.
Right now the master is running 4 concurrent jobs (out of
4) while both
slaves are perfectly idle.
To make the change, simply log into hudson.eclipse.org and
on your job,
click "Tie this project to a node" and select one of the
slaves.
Thanks a bunch,
Denis
--
Eclipse Summit Europe, Nov 2-4
http://www.eclipsecon.org/summiteurope2010/
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
------ End of Forwarded Message
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
|