parrot_cms_setup

#!/bin/sh

glidein_config="$1"

function info {
    echo `date` parrot_cms_setup: $@ 1>&2
}

function warn {
    echo `date` parrot_cms_setup: $@ 1>&2
}

function die {
    errorType="$1"
    shift
    warn ERROR: "$@"
    "$error_gen" -error parrot_cms_setup "$errorType" "$@"
    exit 1
}

function get_squid_version {
    # expected format of Via line in output of wget:
    # Via: 1.1 vocms126.cern.ch:8000 (squid/2.7.STABLE9+fix2831+2833), 1.0 cayenne.hep.wisc.edu:3128 (squid/2.7.STABLE7)
    # Before the comma is the upstream squid.  After is the site squid.
    # At a site where the squid does not cache, I saw the following instead:
    # Via: 1.1 vocms63.cern.ch:8000 (squid/2.7.STABLE9+fix2831+2833)

    env http_proxy="$1" wget -qdO/dev/null --header="X-Frontier-Id: AAA opportunistic probe" "http://cmsfrontier.cern.ch:8000/FrontierProd/Frontier/type=frontier_request:1:DEFAULT&encoding=BLOBzip5&p1=eNorTs1JTS5RUM8vKMgvKinNyywuyUyOLyjKT0pVV0grys9VSClNzAEAIY0Oow__" 2>&1|grep ^Via: | awk -F, '{print $2}' | sed "s|.*(squid/\([^(]*\))[ \r\n]*|\1|"
}

###########################################################
# import add_config_line and add_condor_vars_line functions

add_config_line_source=`grep '^ADD_CONFIG_LINE_SOURCE ' $glidein_config | awk '{print $2}'`
source $add_config_line_source

condor_vars_file=`grep -i "^CONDOR_VARS_FILE " $glidein_config | awk '{print $2}'`

# find error reporting helper script
error_gen=`grep '^ERROR_GEN_PATH ' $glidein_config | awk '{print $2}'`

parrot_cfg=`grep -i "^PARROT_CFG " $glidein_config | awk '{print $2}'`
if [ -z "$parrot_cfg" ]; then
    die VO_Config "failed to find PARROT_CFG in glidein_config"
fi
. $parrot_cfg

# locate the path to the untarred cms_siteconf.tgz

# this case is for <file> in the top-level scope of the frontend config
siteconf_dir=`grep -i "^GLIDECLIENT_GLIDEIN_CMS_SITECONF " $glidein_config | awk '{print $2}'`
if [ -z "$siteconf_dir" ]; then
    # this case is for <file> in the group scope of the frontend config
    siteconf_dir=`grep -i "^GLIDECLIENT_GROUP_GLIDEIN_CMS_SITECONF " $glidein_config | awk '{print $2}'`
fi
if [ -z "$siteconf_dir" ]; then
    # this case is for <file> in the factory config
    siteconf_dir=`grep -i "^GLIDEIN_CMS_SITECONF " $glidein_config | awk '{print $2}'`
fi
if [ -z "$siteconf_dir" ]; then
    die VO_Config "failed to find GLIDECLIENT_GLIDEIN_CMS_SITECONF in glidein_config"
fi

# GLIDECLIENT_GLIDEIN_CMS_SITECONF is a directory containing untarred
# files.  Point siteconf_dir to the SITECONF directory contained
# within the tar file:
info "searching $siteconf_dir for directories"

dirs_found=0
for dir in "$siteconf_dir"/*; do
    if [ -d "$dir/JobConfig" ]; then
        #siteconf_dir="$dir"
        dirs_found=$(($dirs_found+1))
    fi
done

if [ $dirs_found -ne 1 ]; then
    die Corruption "Found $dirs_found directories in the SITECONF tar file: $(echo $siteconf_dir/*)"
fi

info "Replacing SITECONF_PATH_MACRO with $siteconf_dir in site-local-config.xml"

mv "$siteconf_dir/local/JobConfig/site-local-config.xml" \
   "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
 || die WN_Resource "Failed to move site-local-config.xml to replace SITECONF_PATH_MACRO"

sed < "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
    > "$siteconf_dir/local/JobConfig/site-local-config.xml" \
    's|\(SITECONF_PATH_MACRO\)|'"$siteconf_dir"'|' \
  || die WN_Resource "Failed to replace site-local-config.xml to replace SITECONF_PATH_MACRO"


# Test to see if local SQUID proxy is suitable for CMS Frontier.
# See https://twiki.cern.ch/twiki/bin/view/CMS/MyOwnSquid
# This procedure was recommeded by Dave Dykstra.  We require
#   2.6 <= squid_version < 3
# We used to require
#   2.7.STABLE7 <= squid_version < 3
# but Dave decided it was okay for opportunistic usage to allow 2.6 too.

http_proxy=`grep -i "^PROXY_URL " $glidein_config | awk '{print $2}'`
if [ "$http_proxy" = "None" ]; then
    http_proxy=""
fi
if [ "$http_proxy" != "" ] && ! (echo "$http_proxy" | grep -q ^http://); then
    http_proxy="http://$http_proxy"
fi


squid_compatible=0
if [ "$http_proxy" != "" ]; then
    squid_version=$(get_squid_version $http_proxy)

    squid_major=$(echo "$squid_version" | sed -e 's|frontier-squid-||' | awk -F. '{print $1}')
    squid_minor=$(echo "$squid_version" | awk -F. '{print $2}')
    squid_micro=$(echo "$squid_version" | awk -F. '{print $3}')

    echo "Squid major: $squid_major; squid minor: $squid_minor"
    if [ "$squid_major" = 2 ]; then
        if [ "$squid_minor" -ge 6 ]; then
            squid_compatible=1
        fi
    fi

    # publish the squid version (to aid debugging)
    warn "SiteSquidVersion=$squid_version"
    add_config_line SiteSquidVersion "$squid_version"
    add_condor_vars_line SiteSquidVersion "S" "-" "+" "N" "Y" "-"
fi

warn "squid version: $squid_version, proxy $http_proxy (Frontier compatible: $squid_compatible)"

# publish compatibility of the site squid (to aid debugging)
if [ "$squid_compatible" = 1 ]; then
    add_condor_vars_line CMSFrontierUsingSiteSquid "C" "True" "+" "N" "Y" "-"
    add_condor_vars_line HasCMSFrontier "C" "True" "+" "N" "Y" "-"
else
    add_condor_vars_line CMSFrontierUsingSiteSquid "C" "False" "+" "N" "Y" "-"
fi

if [ "$squid_compatible" = 1 ]; then
    # Edit site-local-config.xml.
    # We change two things:
    #  1. insert the local site squid server at the front of the proxy list
    #  2. turn off proxy load-balancing, so we always try local proxy first
    #     before falling back to our central proxies
    #     (TODO: would like to load-balance across central proxies if we
    #      fall back to them; perhaps shuffle the list now)
    #  ASSUMPTION: the original site-local-config.xml contains the
    #  directive <load balance="proxies"/>, and this comes BEFORE
    #  the central proxies.

    warn "Configuring Frontier to use site squid proxy $http_proxy ($squid_version)"

    mv "$siteconf_dir/local/JobConfig/site-local-config.xml" \
        "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
        || die WN_Resource "Failed to move site-local-config.xml"

    sed < "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
        > "$siteconf_dir/local/JobConfig/site-local-config.xml" \
        's|<load balance="proxies"/>|<proxy url="'"$http_proxy"'"/>|' \
        || die WN_Resource "Failed to replace site-local-config.xml"
fi

if [ "$squid_compatible" != 1 ]; then
    # We will have to use the central frontier proxies,
    # so test to make sure they are accessible.

    central_proxies=$(grep 'proxy url=' $siteconf_dir/local/JobConfig/site-local-config.xml | sed 's|.*"\(.*\)".*|\1|' )
    proxy_ok=0
    for http_proxy in $central_proxies; do
        warn "testing central frontier proxy $http_proxy"
        squid_version=$(get_squid_version $http_proxy)
        if [ "$squid_version" = "" ]; then
            warn "FAILED test of http proxy $http_proxy"
        else
            warn "succeeded in test of http proxy $http_proxy: $squid_version"
            proxy_ok=1
        fi
    done
    if [ "$proxy_ok" = 1 ]; then
        add_condor_vars_line HasCMSFrontier "C" "True" "+" "N" "Y" "-"
        warn "Not configuring Frontier to use site squid proxy $http_proxy ($squid_version), because it is not compatible."
    else
        if [ "$GlideinRequiresCMSFrontier" != "" ] && [ "$GlideinRequiresCMSFrontier" != "false" ]; then
            die Network "could not successfully use the site or central frontier proxies: $central_proxies"
        else
            warn "Frontier failed, but GlideinRequiresCMSFrontier is not True, so continuing anyway."

            # Do not let jobs requiring CMS Frontier run here
            GLIDECLIENT_Start=`grep -i "^GLIDECLIENT_Start " $glidein_config | cut -d " " -f 2-`
            if [ "$GLIDECLIENT_Start" != "" ]; then
                GLIDECLIENT_Start="($GLIDECLIENT_Start) && "
            fi
            GLIDECLIENT_Start="${GLIDECLIENT_Start}TARGET.RequiresCMSFrontier=!=True"
            add_config_line GLIDECLIENT_Start "$GLIDECLIENT_Start"
        fi
    fi
fi

CVMFS_OSG_APP=`grep -i "^CVMFS_OSG_APP " $glidein_config | awk '{print $2}'`
if [ "$CVMFS_OSG_APP" != "" ]; then
    export CVMFS_VO_CMS_SW_DIR=$CVMFS_OSG_APP/cmssoft/cms
else
    export CVMFS_VO_CMS_SW_DIR=/cvmfs/cms.cern.ch
fi

# put VO_CMS_SW_DIR into the job environment
add_config_line VO_CMS_SW_DIR $CVMFS_VO_CMS_SW_DIR
add_condor_vars_line VO_CMS_SW_DIR "S" "-" "+" "N" "N" "+"

# append to GLIDEIN_PARROT_OPTIONS
#   - Tell parrot to remap requests for SITECONF from cvmfs to the local SITECONF
#     that ships with this glidein (and which we modified above).
#   ASSUMPTION: the full path to SITECONF is $CVMFS_VO_CMS_SW_DIR/SITECONF

#   Rather than making the above assumption, we would rather use the trick of
#   setting an environment variable (CMS_LOCAL_SITE) that controls the symlink
#   SITECONF/local.  However, parrot currently does not fully support symlinks
#   in CVMFS that point outside the CVMFS repository.
#
#   In addition to this mapping, also replace SITECONF_PATH_MACRO in any
#   existing GLIDEIN_PARROT_OPTIONS with the actual SITECONF path.


GLIDEIN_PARROT=`grep -i "^GLIDEIN_PARROT " $glidein_config | awk '{print $2}'`

PARROT_DEBUG=`grep -i "^PARROT_DEBUG " $glidein_config | awk '{print $2}'`

if [ "X$PARROT_DEBUG" = "" ]; then
  GLIDEIN_PARROT_OPTIONS="$GLIDEIN_PARROT_OPTIONS -d cvmfs"
fi

echo "GLIDEIN_PARROT_OPTIONS=\" \${GLIDEIN_PARROT_OPTIONS/SITECONF_PATH_MACRO/${siteconf_dir}} -M $CVMFS_VO_CMS_SW_DIR/SITECONF=$siteconf_dir\"" \
  >> $GLIDEIN_PARROT/setup.sh \
  || die VO_Config "failed to append to parrot/setup.sh"


# If we are running at a CMS site, fix up the site name in our
# SITECONF to match the local site.
site_config_xml="$OSG_APP/cmssoft/cms/SITECONF/local/JobConfig/site-local-config.xml"
if ! [ -e "$site_config_xml" ]; then
    # At non-CMS sites, we may encounter an available CVMFS mount that is not configured.
    if [ -z "$VO_CMS_SW_DIR" ]; then
        VO_CMS_SW_DIR=/cvmfs/cms.cern.ch
    fi
    echo "Forcing use of parrot; missing site-local-config.xml"
    site_config_xml="$VO_CMS_SW_DIR/SITECONF/local/JobConfig/site-local-config.xml"
fi
if [ -e "$site_config_xml" ]; then
    # Extract site name from: <site name="HERE">
    site_name=$( grep "<site name=" "$site_config_xml" | sed 's|.*<site name *= *"\([^"]*\)".*|\1|' | head -1 )

    if [ "$site_name" != "" ]; then
        info "Inserting site name=$site_name into site-local-config.xml"

        mv "$siteconf_dir/local/JobConfig/site-local-config.xml" \
            "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
        || die VO_Config "Failed to move site-local-config.xml to insert site name"

        sed < "$siteconf_dir/local/JobConfig/site-local-config.xml.orig" \
            > "$siteconf_dir/local/JobConfig/site-local-config.xml" \
            's|<site name="\([^"]*\)"|<site name="'"$site_name"'"|' \
        || die VO_Config "Failed to replace site-local-config.xml to insert site name"
    fi
else
    # We don't have a working site-local-config.xml; force use of parrot.
    touch $GLIDEIN_PARROT/FORCE_PARROT
fi

# If we are running at a CMS site, fix up the xrootd source string
# to match the local site.
storage_xml="$OSG_APP/cmssoft/cms/SITECONF/local/PhEDEx/storage.xml"
if ! [ -e "$storage_xml" ]; then
    # At non-CMS sites, we may encounter an available CVMFS mount that is not configured.
    if [ -z "$VO_CMS_SW_DIR" ]; then
        VO_CMS_SW_DIR=/cvmfs/cms.cern.ch
    fi
    storage_xml="$VO_CMS_SW_DIR/SITECONF/local/PhEDEx/storage.xml"
fi
if [ -e "$storage_xml" ]; then
    # Extract site name from: root://cmsxrootd.fnal.gov//store/$1?source=HERE
    xrootd_source=$( grep "root://cmsxrootd.fnal.gov" "$storage_xml" | sed 's|.*source=\([^"^ ]*\)".*|\1|' | head -1 )

    if [ "$xrootd_source" != "" ]; then
        info "Inserting xrootd source=$xrootd_source into storage.xml"

        mv "$siteconf_dir/local/PhEDEx/storage.xml" \
            "$siteconf_dir/local/PhEDEx/storage.xml.orig" \
        || die VO_Config "Failed to move storage.xml to insert site name"

        sed < "$siteconf_dir/local/PhEDEx/storage.xml.orig" \
            > "$siteconf_dir/local/PhEDEx/storage.xml" \
            's|\(root://cmsxrootd.fnal.gov.*source=\)\([^"^ ]*\)"|\1'"$xrootd_source"'"|' \
        || die VO_Config "Failed to replace storage.xml to insert site name"
    fi
else
    # We don't have a working CVMFS with storage.xml; force use of parrot.
    echo "Forcing use of parrot; missing storage.xml"
    touch $GLIDEIN_PARROT/FORCE_PARROT
fi

# If parrot is forced (possibly due to siteconf) but not working, bail.
PARROT_RUN_WORKS=`grep -i "^PARROT_RUN_WORKS " $glidein_config | awk '{print $2}'`
if [ -e "$GLIDEIN_PARROT/FORCE_PARROT" -a "X$PARROT_RUN_WORKS" != "XTRUE" ]; then
    die WN_Resource "Parrot is needed for SITECONF, but Parrot is not working."
fi


"$error_gen" -ok parrot_cms_setup