Thank you very much Ken for your help. I am a newie without the knowledge to recode the script. I contacted with Htcondor admin about it. Let me see if I am lucky and they can help. My best regards, Livan Valladares Martell.
On Jan 19, 2015, at 6:24 PM, [email protected] wrote:
Send discuss mailing list submissions to [email protected]
To subscribe or unsubscribe via the World Wide Web, visit https://lists.mpich.org/mailman/listinfo/discuss or, via email, send a message with subject or body 'help' to [email protected]
You can reach the person managing the list at [email protected]
When replying, please edit your Subject line so it is more specific than "Re: Contents of discuss digest..."
Today's Topics:
1. Re: MPICH 2 script to MPICH 3 (Kenneth Raffenetti) 2. error installing mpich-master-v3.2a2-113-g7d59d14d on Linux (Siegmar Gross)
----------------------------------------------------------------------
Message: 1 Date: Mon, 19 Jan 2015 10:04:47 -0600 From: Kenneth Raffenetti <[email protected]> To: <[email protected]> Subject: Re: [mpich-discuss] MPICH 2 script to MPICH 3 Message-ID: <[email protected]> Content-Type: text/plain; charset="utf-8"; format=flowed
Hi Livan,
It should be possible to convert your script to use Hydra. Hydra has 2 main components. The first is the main hydra executable - mpiexec. The other is the hydra_pmi_proxy.
In a normal ssh launched scenario, mpiexec will launch proxies on all nodes being used for the job, and those proxies will launch the mpi processes.
If ssh is not possible in your setup, you should be able to utilize the hydra manual laucher (mpiexec -launcher manual). When you run mpiexec with that parameter, it will output the proxy launch commands that can be used to connect to it. It will then wait for all the proxies to connect. Here's an example:
raffenet@doom:mpich/ $ mpiexec -launcher manual -n 3 -hosts a,b,c /bin/hostname HYDRA_LAUNCH: /sandbox/mpich/i/bin/hydra_pmi_proxy --control-port doom:36311 --rmk user --launcher manual --demux poll --pgid 0 --retries 10 --usize -2 --proxy-id 0 HYDRA_LAUNCH: /sandbox/mpich/i/bin/hydra_pmi_proxy --control-port doom:36311 --rmk user --launcher manual --demux poll --pgid 0 --retries 10 --usize -2 --proxy-id 1 HYDRA_LAUNCH: /sandbox/mpich/i/bin/hydra_pmi_proxy --control-port doom:36311 --rmk user --launcher manual --demux poll --pgid 0 --retries 10 --usize -2 --proxy-id 2 HYDRA_LAUNCH_END
Let me know if you have other questions.
Ken
On 01/19/2015 05:39 AM, Livan Valladares wrote:
Hello, I have been using Htcondor with MPICH 2 but I update MPICH 2 to MPICH 3 and now the MPD process manager has been deprecated and Hydra is de default process manager. The script I was using use some commands like mpdtrace, mpdallexit but they are not supported anymore. Here is my script, is it any possibility to change this script code to be able to work with Hydra? Thank you very much, Livan Valladares Martell Script:
#!/bin/sh
##************************************************************** ## ## Copyright (C) 1990-2014, Condor Team, Computer Sciences Department, ## University of Wisconsin-Madison, WI. ## ## Licensed under the Apache License, Version 2.0 (the "License"); you ## may not use this file except in compliance with the License. You may ## obtain a copy of the License at ## ## http://www.apache.org/licenses/LICENSE-2.0 ## ## Unless required by applicable law or agreed to in writing, software ## distributed under the License is distributed on an "AS IS" BASIS, ## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. ## See the License for the specific language governing permissions and ## limitations under the License. ## ##**************************************************************
# Set this to the bin directory of MPICH installation MPDIR=/opt/local/bin PATH=$MPDIR:.:$PATH export PATH
_CONDOR_PROCNO=$_CONDOR_PROCNO _CONDOR_NPROCS=$_CONDOR_NPROCS
# Remove the contact file, so if we are held and released # it can be recreated anew
rm -f $CONDOR_CONTACT_FILE
PATH=`condor_config_val libexec`/:$PATH
# mpd needs a conf file, and it must be # permissions 0700 mkdir tmp MPD_CONF_FILE=`pwd`/tmp/mpd_conf_file export MPD_CONF_FILE
ulimit -c 0
# If you have a shared file system, maybe you # want to put the mpd.conf file in your home # directory
echo "password=somepassword" > $MPD_CONF_FILE chmod 0700 $MPD_CONF_FILE
# If on the head node, start mpd, get the port and host, # and condor_chirp it back into the ClassAd # so the non-head nodes can find the head node.
if [ $_CONDOR_PROCNO -eq 0 ] then mpd > mpd.out.$_CONDOR_PROCNO 2>&1 & sleep 1 host=`mpdtrace -l | sed 1q | tr '_' ' ' | awk '{print $1}'` port=`mpdtrace -l | sed 1q | tr '_' ' ' | awk '{print $2}'`
condor_chirp set_job_attr MPICH_PORT $port condor_chirp set_job_attr MPICH_HOST \"$host\"
num_hosts=1 retries=0 while [ $num_hosts -ne $_CONDOR_NPROCS ] do num_hosts=`mpdtrace | wc -l` sleep 2 retries=`expr $retries + 1` if [ $retries -gt 100 ] then echo "Too many retries, could not start all $_CONDOR_NPROCS nodes, only started $num_hosts, giving up. Here are the hosts I could start " mpdtrace exit 1 fi done
## run the actual mpi job, which was the command line argument ## to the invocation of this shell script mpiexec -n $_CONDOR_NPROCS $@ e=$?
mpdallexit sleep 20 echo $e else # If NOT the head node, acquire the host and port of # the head node retries=0 host=UNDEFINED while [ $host == "UNDEFINED" ] do host=`condor_chirp get_job_attr MPICH_HOST` sleep 2 retries=`expr $retries + 1` if [ $retries -gt 100 ]; then echo "Too many retries, could not get mpd host from condor_chirp, giving up." exit 1 fi done
port=`condor_chirp get_job_attr MPICH_PORT` host=`echo $host | tr -d '"'` mpd --host=$host --port=$port > mpd.out.$_CONDOR_PROCNO 2>&1 fi
_______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss
------------------------------
Message: 2 Date: Mon, 19 Jan 2015 17:24:22 +0100 (CET) From: Siegmar Gross <[email protected]> To: [email protected] Subject: [mpich-discuss] error installing mpich-master-v3.2a2-113-g7d59d14d on Linux Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii"
Hi,
today I tried to install mpich-master-v3.2a2-113-g7d59d14d on my machines (Solaris 10 Sparc, Solaris 10 x86_64, and openSUSE Linux 12.1 x86_64) with gcc-4.9.2 and Sun C 5.13. I succedded on both Solaris machines with both compilers and got the following error on Linux for both compilers.
tyr mpich-master-v3.2a2-113-g7d59d14d-Linux.x86_64.64_gcc 304 cat log.make.Linux.x86_64.64_gcc if test ! -h ./src/include/mpio.h ; then \ rm -f ./src/include/mpio.h ; \ ( cd ./src/include && \ ln -s ../mpi/romio/include/mpio.h ) ; \ fi make all-recursive make[1]: Entering directory `/export2/src/mpich-3.2/mpich-master-v3.2a2-113-g7d59d14d-Linux.x86_64.64_gcc' make[1]: execvp: /bin/sh: Argument list too long make[1]: *** [all-recursive] Error 127 make[1]: Leaving directory `/export2/src/mpich-3.2/mpich-master-v3.2a2-113-g7d59d14d-Linux.x86_64.64_gcc' make: *** [all] Error 2 tyr mpich-master-v3.2a2-113-g7d59d14d-Linux.x86_64.64_gcc 305
I used the following "configure" command.
../mpich-master-v3.2a2-113-g7d59d14d/configure --prefix=/usr/local/mpich-3.2_64_gcc \ --libdir=/usr/local/mpich-3.2_64_gcc/lib64 \ --includedir=/usr/local/mpich-3.2_64_gcc/include64 \ CC="gcc" CXX="g++" F77="gfortran" FC="gfortran" \ CFLAGS="-m64" CXXFLAGS="-m64" FFLAGS="-m64" FCFLAGS="-m64" \ LDFLAGS="-m64 -L/usr/lib/sparcv9 -Wl,-rpath -Wl,/usr/lib/sparcv9" \ --enable-f77 --enable-fc --enable-cxx --enable-romio \ --enable-debuginfo --enable-smpcoll \ --enable-threads=runtime --with-thread-package=posix \ --enable-shared \ |& tee log.configure.$SYSTEM_ENV.$MACHINE_ENV.64_gcc
I would be grateful if somebody can fix the problem. Thank you very much for any help in advance.
Kind regards
Siegmar