Warning: Undefined array key 16 in /usr/share/dokuwiki/inc/html.php on line 1453

Differences

This shows you the differences between two versions of the page.

--- cluster:146 [2016/02/10 13:07]
127.0.0.1 external edit
+++ cluster:146 [2017/08/29 09:36] (current)
hmeij07
@@ Line 2: / Line 2: @@
 **[[cluster:0|Back]]**
-==== Brief Guide to HPCC ====
+==== Openlava 3.1.2 ====
-This page will be maintained and provide information to get users started using the compute clusters. It is a merger of the old "brief description" page and the "queue description" page.
+Build process, switching to ''git'' approach.
-===== Description =====
+Prequisites (for rpm.sh)
+  * yum install git
+  * yum install rpm-build
+  * yum install rpmdevtools
+  * yum install tcl tcl-devel
+  * yum install ncurses ncurses-devel
+  * yum install automake libtool
+<code>
+# initialize git
+git init
+mkdir git-latest
+cd git-latest
+# pull branch down
+git clone git://github.com/openlava/openlava.git
+cd openlava/
+# what's new
+git log
+# pull Dave's addition of rpm.sh
+git checkout 07f740f884d9c71d
+# build
+./bootstrap.sh
+./configure --prefix=/opt/openlava-3.1.2
+make
+# build packages
+./rpm.sh
+find/root/rpmbuild/ -name *.rpm
+</code>
+Next we remove a queue and its compute nodes from the current openlava configuration and restart those services.  Then we configure the new openlava config files. Once we have a few nodes up and one claims to be the master we do a manual install until we have enough to build a script. For my memory sake...
+<code>
+#!/bin/bash
+cd /share/apps/src/petaltail6/openlava3/
+service openlava stop
+chkconfig openlava off
+# remove and clean, use rpm -ql list too
+yum erase openlava -y --disablerepo=*
+rm -rf /opt/openlava-2.2*
+rm -rf /var/spool/mail/openlava
+# not needed # ./openlava-2.2.rmpql
+# still there # grep openlava /etc/passwd
+foo=`rpm -qa | grep tcl-devel | wc -l`
+if [ $foo -eq 1 ]; then
+        echo "tcl-devel installed"
+else
+        echo "tcl-devel installing"
+        yum localinstall --nogpgcheck -y tcl-devel-8.5.7-6.el6.x86_64.rpm
+fi
+yum localinstall --disablerepo=* --nogpgcheck -y ./rpmbuild6_jan16/RPMS/x86_64/openlava-3.1-1.x86_64.rpm
+scp /etc/init.d/openlava /etc/init.d/openlava-orig
+scp ./openlava.ulimit /etc/init.d/openlava
+scp /share/apps/scripts/elim.gpu /opt/openlava-3.1/sbin/elim
+foo=`grep 'ulimit -l' /etc/pam.d/sshd | wc -l`
+if [ $foo -eq 1 ]; then
+        echo "ulimit code present in /etc/pam.d/sshd"
+else
+        echo "ulimit code added to /etc/pam.d/sshd"
+        echo -e "# for ulimit -l -hmeij\nsession    required     pam_limits.so\n" >> /etc/pam.d/sshd
+fi
+foo=`grep 'libibverbs' /etc/security/limits.conf | wc -l`
+if [ $foo -eq 1 ]; then
+        echo "ulimit code present in /etc/security/limits.conf"
+else
+        echo "ulimit code added to /etc/security/limits.conf"
+        echo -e "# added for RLIMIT_MEMLOCK warnings with libibverbs -hmeij\n*                soft    memlock         unlimited\n*                hard    memlock         unlimited\n" >> /etc/security/limits.conf
+fi
+scp /etc/postfix/main.cf /etc/postfix/main.cf-orig
+scp ./etc_postfix_main.cf /etc/postfix/main.cf
+chkconfig postfix on
+chkconfig openlava --list
+chkconfig postfix --list
+#etc
+mv /opt/openlava-3.1/etc /opt/openlava-3.1/etc-orig
+scp -r ./etc /opt/openlava-3.1/
+#last for usage reporting
+ln -s /opt/openlava-3.1 /opt/openlava-2.2
+chown -R openlava:openlava /opt/openlava-3.1
+echo reboot
+reboot
+</code>
+and now deploy from command line
+<code>
+for i in `seq 4 32`; do echo n$i; ssh n$i /share/apps/src/petaltail6/openlava3/cmd.sh; done
+</code>
+During the final roll out, copy the /opt/openlava-2.2/work directory to /opt/openlava-3.1 directory. Remove work/logdir/#lsb.events.lock and all files in work/logdir/info to preserve stats.
-The High Performance Compute Cluster (HPCC) is comprised of two login nodes (greentail.wesleyan.edu and swallowtail.wesleyan.edu, both Dell PowerEdge 2950s).  Old login node petaltail.wesleyan.edu (Dell PowerEdge 2950) can be used for testing code (does not matter if it crashes, it's primary duty is backup to virtual tape library).
 \\
 **[[cluster:0|Back]]**

DokuWiki

User Tools

Site Tools

Differences

Page Tools