EntryDescriptionNo52 < Main

Tags: , view all tags
---+ Upgrade to OSG 1.0
We started our upgrade to 1.0:
   * stoping our gatekeeper, in our computing element
<pre>
/etc/init.d/xinetd stop
</pre>
   * In our Computing element, getting pacman
<pre>
cp -a  /opt/osg-0.8.0 /home/mdias/osg_backup0.8
cd /opt
wget http://physics.bu.edu/pacman/sample_cache/tarballs/pacman-3.26.tar.gz
tar --no-same-owner -xzvf pacman-3.26.tar.gz
cd pacman-3.26
source setup.sh
</pre>
now, 
<pre>
cd /opt
mkdir osg-1.0.0
cd /opt/osg-1.0.0
vdt-control --off
disabling init service tomcat-55... ok
disabling cron service gratia-condor... ok
disabling init service condor... ok
disabling init service syslog-ng... ok
disabling init service tomcat-5... ok
disabling init service osg-rsv... ok
disabling init service apache... ok
disabling init service condor-devel... ok
disabling cron service vdt-update-certs... ok
disabling init service MLD... ok
disabling cron service gums-host-cron... ok
disabling cron service edg-mkgridmap... ok
disabling init service globus-ws... ok
disabling init service mysql... ok
disabling inetd service gsiftp... ok
disabling inetd service globus-gatekeeper... ok
disabling init service gris... ok
disabling cron service vdt-rotate-logs... ok
disabling cron service fetch-crl... ok

export OLD_VDT_LOCATION=/opt/osg-0.8.0
pacman -get OSG:ce
source setup.sh
pacman -get OSG:Globus-Condor-Setup
</pre>
the last step install condor batch. We have to inspect
<pre>
 vim  /opt/osg-1.0.0/condor/etc/condor_config
</pre>
Managed fork issues: installing and configuring it 
<pre>
pacman -get  OSG:ManagedFork
 $VDT_LOCATION/vdt/setup/configure_globus_gatekeeper --managed-fork y --server y
</pre>
Gums server:
<pre>
pacman -get  OSG:gums
</pre>
Monalisa
<pre>
 vim /opt/osg-1.0.0/MonaLisa/Service/VDTFarm/ml.properties
$VDT_LOCATION/vdt/setup/configure_monalisa --prompt
vim  /opt/osg-1.0.0/MonaLisa/Service/CMD/ml_env
JAVA_HOME=/opt/osg-1.0.0/jdk1.5  <-------was jdk1.4
FARM_NAME=SPRACE
</pre>
(Look this after services are on. It was changed.)
Gums host cron
<pre>
vdt-register-service -name gums-host-cron --enable
vdt-register-service: updated cron service 'gums-host-cron'
vdt-register-service: desired state = enable
vdt-register-service: cron time     = '9 8,14,20,2 * * *'
vdt-register-service: cron command  = '/opt/osg-1.0.0/gums/scripts/gums-host-cron'

 vdt-control --enable gums-host-cron
$VDT_LOCATION/gums/scripts/gums-host-cron
</pre>
some gums configuration
<pre>
cp /opt/osg-1.0.0/post-install/prima-authz.conf /etc/grid-security/.
cp /opt/osg-1.0.0/post-install/gsi-authz.conf /etc/grid-security/.
/opt/osg-1.0.0/tomcat/v55/webapps/gums/WEB-INF/scripts/gums-add-mysql-admin "/DC=org/DC=doegrids/OU=People/CN=Marco Dias 280904"
</pre>
We have to inspect these files
<pre>
$VDT_LOCATION/vdt-app-data/gums/config/gums.config
$VDT_LOCATION/gums/config/gums-client.properties
</pre>

Reusing the old configuration
<pre>
cd /opt/osg-1.0.0/monitoring
source ../setup.sh
export OLD_VDT_LOCATION=/opt/osg-0.8.0/
./configure-osg.py -e
vim extracted-config.ini
</pre>
we can test it and repair
<pre>
./configure-osg.py -v -f ./extracted-config.ini
 cp /opt/osg-0.8.0/monitoring/grid3-user-vo-map.txt
/opt/osg-1.0.0/monitoring/.
vim /opt/osg-1.0.0/monitoring/extracted-config.ini
</pre>
installing it.
<pre>
  ./configure-osg.py -c -f ./extracted-config.ini 
</pre>
removing and creating simbolic links
<pre>
clcmd umount /OSG/
unlink /OSG 
ln -s /opt/osg-1.0.0 /OSG
vim /etc/exports
 /etc/init.d/nfs restart
</pre>
Set  Globus-Base-WSGRAM-Server
<pre>
visudo
/opt/osg-1.0.0/vdt/setup/configure_prima_gt4 --enable --gums-server osg-ce.sprace.org.br
/etc/init.d/globus-ws stop
/etc/init.d/globus-ws start
</pre>

Condor-cron
<pre>
vim /opt/osg-1.0.0/condor-cron/local.osg-ce/condor_config.local
chown condor: /var/lock/subsys/condor-cron/
</pre>
we also add in our 
<pre>
vim $VDT_LOCATION/condor-cron/etc/condor_config file
NETWORK_INTERFACE = 192.168.1.150
</pre>

Turning on services 
<pre>
vdt-control --on
enabling cron service fetch-crl... ok
enabling cron service vdt-rotate-logs... ok
enabling cron service vdt-update-certs... ok
skipping init service 'gris' -- marked as disabled
enabling inetd service globus-gatekeeper... ok
enabling inetd service gsiftp... ok
enabling init service mysql... ok
enabling init service globus-ws... ok
skipping cron service 'edg-mkgridmap' -- marked as disabled
enabling cron service gums-host-cron... ok
enabling init service MLD... ok
enabling init service condor-cron... ok
enabling init service apache... ok
skipping init service 'osg-rsv' -- marked as disabled
enabling init service tomcat-55... ok
enabling init service syslog-ng-sender... ok
enabling init service condor... ok
enabling cron service gratia-condor... ok

</pre>

Configurating CEmon and seeing if GIP is working
<pre>
$VDT_LOCATION/vdt/setup/configure_cemon --consumer https://osg-ress-1.fnal.gov:8443/ig/services/CEInfoCollector --topic OSG_CE
$VDT_LOCATION/verify-gip-for-cemon/verify-gip-for-cemon-wrapper
</pre>

Some notes: gums used the old configuration, but gets the old password also, in your gums.config
<pre>
 hibernate.connection.password=
</pre>
so we had to inspect the vdt-install.log, looking where did it set the password and insert in that field. 

An issue with condor: condor_status ok, but condor_q freezes. We suspected a port conflict with condor-cron. We configurated:
<pre>
COLLECTOR_HOST  = $(CONDOR_HOST):9619 #was 9618
HIGHPORT = 65100 #commentaded 9700
LOWPORT = 65001 #commented 9600
CREDD_PORT                      = 9622 #was 9620
STORK_PORT                      = 9623 #was 9621
</pre>
and in our condor
<pre>
HIGHPORT = 65000
LOWPORT = 63001
</pre>
but the real problem was a bug in 7.0.2 version. We installed again
<pre>
cd /OSG/
cp -a condor condor.old
cd condor
 rm -rf *
cd ..
wget
http://parrot.cs.wisc.edu//symlink/20080702041503/7/7.0/7.0.3/19811b56762d4e4ed3ea991b9a341232/condor-7.0.3-linux-x86-rhel3.tar.gz
tar -xvzf condor-7.0.3-linux-x86-rhel3.tar.gz
 cd condor-7.0.3
./condor_configure --install --maybe-daemon-owner --make-personal-condor
--install-log ../post-install/README --install-dir /OSG/condor
 cd /OSG/condor/etc/
 mv condor_config condor_config.bck
cp /OSG/condor.old/etc/condor_config .

</pre>
   * Installing RSV
<pre>
cd /opt/osg-1.0.0/monitoring/
vim extracted-config.ini
[RSV]
enabled = True
rsv_user = mdias
enable_ce_probes = True
ce_hosts = osg-ce.sprace.org.br
enable_gridftp_probes = True
gridftp_dir = /tmp
enable_gums_probes =  False
gums_hosts = osg-ce.sprace.org.br
enable_srm_probes = True
srm_hosts = osg-se.sprace.org.br
srm_dir = /pnfs/sprace.org.br/data/mdias
use_service_cert = False
proxy_file = /tmp/x509up_u537
enable_gratia = True
print_local_time = True
setup_rsv_nagios = False
setup_for_apache = True
</pre>
now start services
<pre>
 vdt-control --off
./configure-osg.py -c -f ./extracted-config.ini 
vdt-control --on 
</pre>
and check it
<pre>
condor_cron_q
tail -f $VDT_LOCATION/osg-rsv/logs/consumers/gratia-script-consumer.out
</pre>
also looking at https://osg-ce.sprace.org.br:8443/rsv

   * In our nodes, stop condor
<pre>
clcmd /etc/init.d/condor stop
clcmd mount /OSG
</pre>
 in a NFS shared directory,
<pre>
cd /home/mdias
wget http://physics.bu.edu/pacman/sample_cache/tarballs/pacman-3.26.tar.gz
tar --no-same-owner -xzvf pacman-3.26.tar.gz
</pre>
and then  We've created a script like that:
<pre>
source /opt/OSG-wn-client/setup.sh
vdt-control --off
cd /home/mdias/pacman-3.26
source setup.sh
mv /opt/OSG-wn-client /opt/OSG-wn-client.old
mkdir /opt/OSG-wn-client
cd /opt/OSG-wn-client
VDTSETUP_AGREE_TO_LICENSES=y
export VDTSETUP_AGREE_TO_LICENSES
VDTSETUP_INSTALL_CERTS=l
export VDTSETUP_INSTALL_CERTS
VDTSETUP_EDG_CRL_UPDATE=n
export VDTSETUP_EDG_CRL_UPDATE
VDTSETUP_ENABLE_ROTATE=y
export VDTSETUP_ENABLE_ROTATE
VDTSETUP_CA_CERT_UPDATER=n
export VDTSETUP_CA_CERT_UPDATER
pacman -trust-all-caches -get OSG:wn-client
mv /var/log/glexec /var/log/glexec.old
mv  /etc/glexec  /etc/glexec.old
mkdir /opt/glexec
cd /opt/glexec
pacman -trust-all-caches    -get http://vdt.cs.wisc.edu/vdt_181_cache:Glexec
sed -i 's/yourmachine.yourdomain/osg-ce.sprace.org.br/g'  /etc/glexec/contrib/gums_interface/getmapping.cfg
source setup.sh
vdt-control --on
mkdir /opt/glexec/glite/etc
cp /OSG/glite/etc/vomses /opt/glexec/glite/etc/.
</pre>
so, We start to run it in our nodes:
<pre>
/root/bin/clcmd /home/mdias/worknodeinstall.sh
/root/bin/clcmd rm -rf /opt/OSG-wn-client.old
</pre>
we have to edit one /etc/glexec/glexec.conf to put "linger =no" line in [glexec] section and copy to the other nodes.
<pre>
clcmd cp -f /home/mdias/glexec.conf /etc/glexec/.
</pre>
 Also we have to edit our computer element , *before* reusing old configuration:
<pre>
 vim /opt/osg-1.0.0/monitoring/extracted-config.in
glexec_location = /opt/glexec/glexec-osg
</pre>
Testing glexec
<pre>
cd /etc/glexec
source setup.sh 
voms-proxy-init -voms cms:/cms
export GLEXEC_CLIENT_CERT=/tmp/x509up_u537  
cd /opt/glexec/
glexec-osg/sbin/glexec /usr/bin/id
</pre>

A srm error has forced us to upgrade our dcache to make srm compatible. So in our storage element (osg-se):
<pre>
/opt/init.d/dcache-pool stop
/opt/init.d/dcache-core stop
wget http://www.dcache.org/downloads/1.8.0/dcache-server-1.8.0-15p7.noarch.rpm
wget http://www.dcache.org/downloads/1.8.0/dcache-srmclient-1.8.0-15p7.noarch.rpm
rpm -Uvh dcache-server-1.8.0-15p7.noarch.rpm dcache-srmclient-1.8.0-15p7.noarch.rpm
</pre>
Restarting  pools and dcache-core :
<pre>
/opt/d-cache/install/install.sh
 /opt/d-cache/bin/dcache-core start
/opt/d-cache/bin/dcache-pool start
</pre>
Restarting like that
<pre>
[root@spraid02 ~]# /opt/d-cache/bin/dcache-core start
This script is deprecated and will be removed in a future
release. Please use /opt/d-cache/bin/dcache start instead.
Starting dcache services: 
Starting gridftp-spraid02Domain  Done (pid=1147)
[root@spraid02 ~]# /opt/d-cache/bin/dcache-pool start
This script is deprecated and will be removed in a future
release. Please use /opt/d-cache/bin/dcache start pool instead.
WARNING: the variable DCACHE_HOME is not set.
WARNING: Using deprecated value of DCACHE_BASE_DIR as DCACHE_HOME
start dcache pool: Starting spraid02Domain  Done (pid=1271)

</pre>

   * In our storage elements
<pre>
 unlink /opt/osg-0.8.0
 ln -s /OSG /opt/osg-1.0.0
</pre>

configuring dcache information provider:
<pre>
 yum install postgresql.i386
$GIP_LOCATION/conf/configure_gip_dcache
What is the hostname of your dCache admin interface?
 osg-se.sprace.org.br
Configuration saved.  If you would like to alter any choices without 
re-running this configuration script, you may find these answers in:

$VDT_LOCATION/gip/etc/dcache_storage.conf
 chown daemon:root $VDT_LOCATION/gip/etc/dcache_storage.conf
</pre>

interesting links
   * http://twiki.mwt2.org/bin/view/Main/DeployingOSG1d0
   * https://twiki.grid.iu.edu/twiki/bin/view/ReleaseDocumentation/ComputeElementInstall
---++Updates
---+++Marco at 07/10/2008
Apply a GIP patch
<pre>
--- gip/libexec/services_info_provider.py    (revision 2030)
+++ gip/libexec/services_info_provider.py    (working copy)
@@ -102,6 +102,16 @@
 def print_srm(cp, admin):
     sename = cp.get("se", "unique_name")
     sitename = cp.get("site", "unique_name")
+    # BUGFIX: Resolve the IP address of srm host that the admin specifies.
+    # If this IP address matches the IP address given by dCache, then we will
+    # print out the admin-specified hostname instead of looking it up.  This
+    # is for sites where the SRM host is a CNAME instead of the A name.
+    srm_host = cp_get(cp, "se", "srm_host", None)
+    if srm_host:
+        try:
+            srm_ip = socket.gethostbyname(srm_host)
+        except:
+            srm_ip = None
     #vos = [i.strip() for i in cp.get("vo", "vos").split(',')]
     vos = voListStorage(cp)
     ServiceTemplate = getTemplate("GlueService", "GlueServiceUniqueID")
@@ -121,8 +131,11 @@
         hostname = hostname.split(',')[0]
         try:
             hostname = socket.getfqdn(hostname)
+            hostname_ip = socket.gethostbyname(hostname)
         except:
-            pass
+            hostname_ip = None
+        if hostname_ip != None and hostname_ip == srm_ip and srm_host !=
None:
+            hostname = srm_host
         info = {
                 "serviceType"  : "SRM",
                 "acbr"         : acbr[1:], 

</pre>
one more patch
<pre>
--- vdt/setup/configure_gip    (revision 2047)
+++ vdt/setup/configure_gip    (working copy)
@@ -1337,6 +1337,19 @@

     }

+    # BUGFIX: se_access is not filled in if you are using dynamic_dcache, but
+    # this global variable must be populated for the CESE bind to work right
+    # later on
+    if ($srm && $dynamic_dcache==1) {
+
+        foreach $vo (@vo_list){
+            my $se_access_root = $vo_access_roots{$vo};
+            if ($se_access_root!~/^$/) {
+                $se_access{$vo} = $se_access_root;
+            }
+        }
+    }
+
     my $service_config_file =
"$gip_location/etc/osg-info-gip-config/osg-info-static-service.conf";
     safe_write($service_config_file, $service_contents);
     vdt_install_log("===== BEGIN osg-info-static-service.conf =====\n");
@@ -1536,8 +1549,9 @@
             }

             if ($srm && exists $se_access{$vo} && defined $sa_path) {
+                # BUGFIX: use $se_access{$vo} instead of
$sa_path/$se_access{$vo}.
                 $contents = $contents."\ndn: GlueCESEBindSEUniqueID=$se_host,
GlueCESEBindGroupCEUniqueID=$fullhost:2119/jobmanager-$batch-$vo\n"
-                                     ."GlueCESEBindCEAccesspoint:
                                      $sa_path/$se_access{$vo}\n"
+                                     ."GlueCESEBindCEAccesspoint:
$se_access{$vo}\n"
                                      ."GlueCESEBindCEUniqueID:
$fullhost:2119/jobmanager-$batch-$vo\n";
             }
         }

</pre>
now run configure_gip again. Another patch solves a error runing
$GIP_LOCATION/libexec/token_info_provider.py
<pre>

--- gip/lib/python/gip_cese_bind.py    (revision 2047)
+++ gip/lib/python/gip_cese_bind.py    (working copy)
@@ -55,6 +55,8 @@
     ce_list = getCEList(cp)
     se_list = getSEList(cp)
     access_point = cp_get(cp, "vo", "default", "/")
+    if not access_point:
+        access_point = "/UNAVAILABLE"
     for ce in ce_list:
         for se in se_list:
             info = {'ceUniqueID' : ce, 
</pre>
It's not necessary run configure_gip again
---+++Marco at 07/16/2008
Cemon not reporting to BDII database. Parag Mhashilkar gently helped us with
it
<pre>

 Coincidentally, I have seen similar error in catalina.out. The
 admin their claimed that they managed to fix the problem by
 putting xercesImpl.jar in $VDT_LOCATION/tomcat/v55/common/endorsed.

 The claim is that, this missing jar file is resulting in some
 strange interference between gums and cemon installation. You can
 find this jar at couple of places within vdt itself.

 Once you have this file in above dir, stop tomcat, make sure that
 everything in $GLITE_LOCATION/var/cemonitor is deleted (manually)
 and start tomcat.
</pre>
---+++ Marco at 07/18/2008
in order to our dcache GIP works, we havet to edit:
<pre>
 more /OSG/monitoring/extracted-config.ini 
dynamic_dcache = 1
cd /OSG/monitoring
./configure-osg.py -c -f ./extracted-config.ini
</pre>

-- Main.MarcoAndreFerreiraDias - 23 Jun 2008
Topic revision: r12 - 2008-07-03 - MarcoAndreFerreiraDias
antalya escort bursa escort eskisehir escort istanbul escort izmir escort