#!/bin/bash
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
+
+DEBUG=0
+SSH_PORT=22
+PUPPET_CONCURRENCY=5
+
+read -d] -r SCOPES <<EOF
+--scopes
+'["GET /arvados/v1/virtual_machines",\n
+"GET /arvados/v1/keep_services",\n
+"GET /arvados/v1/keep_services/",\n
+"GET /arvados/v1/groups",\n
+"GET /arvados/v1/groups/",\n
+"GET /arvados/v1/links",\n
+"GET /arvados/v1/collections",\n
+"POST /arvados/v1/collections",\n
+"POST /arvados/v1/links",\n
+"GET /arvados/v1/users/current",\n
+"POST /arvados/v1/users/current",\n
+"GET /arvados/v1/jobs",\n
+"POST /arvados/v1/jobs",\n
+"GET /arvados/v1/pipeline_instances",\n
+"POST /arvados/v1/pipeline_instances",\n
+"PUT /arvados/v1/pipeline_instances/",\n
+"GET /arvados/v1/collections/",\n
+"POST /arvados/v1/collections/",\n
+"GET /arvados/v1/logs"]'
+EOF
-read -rd "\000" helpmessage <<EOF
-$(basename $0): Deploy Arvados to a cluster
-
-Syntax:
- $(basename $0) <identifier>
-
-Options:
-
- identifier Arvados cluster name
+function usage {
+ echo >&2
+ echo >&2 "usage: $0 [options] <identifier>"
+ echo >&2
+ echo >&2 " <identifier> Arvados cluster name"
+ echo >&2
+ echo >&2 "$0 options:"
+ echo >&2 " -p, --port <ssh port> SSH port to use (default 22)"
+ echo >&2 " -c, --concurrency <max> Maximum concurrency for puppet runs (default 5)"
+ echo >&2 " -d, --debug Enable debug output"
+ echo >&2 " -h, --help Display this help and exit"
+ echo >&2
+ echo >&2 "Note: this script requires an arvados token created with these permissions:"
+ echo >&2 ' arv api_client_authorization create_system_auth \'
+ echo -e $SCOPES"]'" >&2
+ echo >&2
+}
-EOF
+# NOTE: This requires GNU getopt (part of the util-linux package on Debian-based distros).
+TEMP=`getopt -o hdp:c: \
+ --long help,debug,port:,concurrency: \
+ -n "$0" -- "$@"`
+
+if [ $? != 0 ] ; then echo "Use -h for help"; exit 1 ; fi
+# Note the quotes around `$TEMP': they are essential!
+eval set -- "$TEMP"
+
+while [ $# -ge 1 ]
+do
+ case $1 in
+ -p | --port)
+ SSH_PORT="$2"; shift 2
+ ;;
+ -c | --concurrency)
+ PUPPET_CONCURRENCY="$2"; shift 2
+ ;;
+ -d | --debug)
+ DEBUG=1
+ shift
+ ;;
+ --)
+ shift
+ break
+ ;;
+ *)
+ usage
+ exit 1
+ ;;
+ esac
+done
IDENTIFIER=$1
if [[ "$IDENTIFIER" == '' ]]; then
- echo >&2 "$helpmessage"
- echo >&2
+ usage
exit 1
fi
COLUMNS=80
+PUPPET_AGENT='
+now() { date +%s; }
+let endtime="$(now) + 600"
+while [ "$endtime" -gt "$(now)" ]; do
+ puppet agent --test --detailed-exitcodes
+ agent_exitcode=$?
+ if [ 0 = "$agent_exitcode" ] || [ 2 = "$agent_exitcode" ]; then
+ break
+ else
+ sleep 10s
+ fi
+done
+exit ${agent_exitcode:-99}
+'
+
title () {
- printf "\n%*s\n\n" $(((${#title}+$COLUMNS)/2)) "********** $1 **********"
+ date=`date +'%Y-%m-%d %H:%M:%S'`
+ printf "$date $1\n"
}
-echo $WORKSPACE
-
function run_puppet() {
node=$1
- return_var=$2
+ title "Running puppet on $node"
+ sleep $[ $RANDOM / 6000 ].$[ $RANDOM / 1000 ]
TMP_FILE=`mktemp`
- ssh -t -p2222 -o "StrictHostKeyChecking no" -o "ConnectTimeout 5" root@$node.$IDENTIFIER -C "/usr/bin/puppet agent -t" | tee $TMP_FILE
+ if [[ "$DEBUG" != "0" ]]; then
+ ssh -t -p$SSH_PORT -o "StrictHostKeyChecking no" -o "ConnectTimeout 5" root@$node -C bash -c "'$PUPPET_AGENT'" | tee $TMP_FILE
+ else
+ ssh -t -p$SSH_PORT -o "StrictHostKeyChecking no" -o "ConnectTimeout 5" root@$node -C bash -c "'$PUPPET_AGENT'" > $TMP_FILE 2>&1
+ fi
- ECODE=$?
+ ECODE=${PIPESTATUS[0]}
RESULT=$(cat $TMP_FILE)
if [[ "$ECODE" != "255" && ! ("$RESULT" =~ 'already in progress') && "$ECODE" != "2" && "$ECODE" != "0" ]]; then
- # Puppet exists 255 if the connection timed out. Just ignore that, it's possible that this node is
- # a compute node that was being shut down.
+ # Ssh exits 255 if the connection timed out. Just ignore that.
# Puppet exits 2 if there are changes. For real!
# Puppet prints 'Notice: Run of Puppet configuration client already in progress' if another puppet process
# was already running
- echo "ERROR updating $node.$IDENTIFIER: exit code $ECODE"
+ echo "ERROR running puppet on $node: exit code $ECODE"
+ if [[ "$DEBUG" == "0" ]]; then
+ title "Command output follows:"
+ echo $RESULT
+ fi
+ fi
+ if [[ "$ECODE" == "255" ]]; then
+ title "Connection timed out"
+ ECODE=0
+ fi
+ if [[ "$ECODE" == "2" ]]; then
+ ECODE=0
fi
- rm -f $TMP_FILE
- echo
- eval "$return_var=$ECODE"
-}
-
-# Deploy API server
-title "Deploying API server"
-SUM_ECODE=0
+ if [[ "$ECODE" == "0" ]]; then
+ rm -f $TMP_FILE
+ echo $node successfully updates
+ else
+ echo $node exit code: $ECODE see $TMP_FILE for details
+ fi
+}
-# Install updated debian packages
-title "Deploying updated arvados debian packages"
+function run_command() {
+ node=$1
+ return_var=$2
+ command=$3
-ssh -p2222 root@$IDENTIFIER.arvadosapi.com -C "apt-get update && apt-get -qqy install arvados-src python-arvados-fuse python-arvados-python-client arvados-api-server-bare"
+ title "Running '$command' on $node"
+ TMP_FILE=`mktemp`
+ if [[ "$DEBUG" != "0" ]]; then
+ ssh -t -p$SSH_PORT -o "StrictHostKeyChecking no" -o "ConnectTimeout 125" root@$node -C "$command" | tee $TMP_FILE
+ else
+ ssh -t -p$SSH_PORT -o "StrictHostKeyChecking no" -o "ConnectTimeout 125" root@$node -C "$command" > $TMP_FILE 2>&1
+ fi
-ECODE=$?
-SUM_ECODE=$(($SUM_ECODE + $ECODE))
+ ECODE=$?
+ RESULT=$(cat $TMP_FILE)
-ssh -p2222 root@$IDENTIFIER.arvadosapi.com -C "/usr/local/bin/arvados-api-server-upgrade.sh"
+ if [[ "$ECODE" != "255" && "$ECODE" != "0" ]]; then
+ # Ssh exists 255 if the connection timed out. Just ignore that, it's possible that this node is
+ # a shell node that is down.
+ title "ERROR running command on $node: exit code $ECODE"
+ if [[ "$DEBUG" == "0" ]]; then
+ title "Command output follows:"
+ echo $RESULT
+ fi
+ fi
+ if [[ "$ECODE" == "255" ]]; then
+ title "Connection timed out"
+ ECODE=0
+ fi
+ rm -f $TMP_FILE
+ eval "$return_var=$ECODE"
+}
-ECODE=$?
+title "Updating API server"
+SUM_ECODE=0
+run_puppet $IDENTIFIER.arvadosapi.com ECODE
SUM_ECODE=$(($SUM_ECODE + $ECODE))
if [[ "$SUM_ECODE" != "0" ]]; then
- title "!!!!!! DEPLOYING DEBIAN PACKAGES FAILED !!!!!!"
+ title "ERROR: Updating API server FAILED"
EXITCODE=$(($EXITCODE + $SUM_ECODE))
exit $EXITCODE
fi
-title "Deploying updated arvados debian packages complete"
-
-# Install updated arvados gems
-title "Deploying updated arvados gems"
-
-ssh -p2222 root@$IDENTIFIER.arvadosapi.com -C "/usr/local/rvm/bin/rvm default do gem install arvados arvados-cli && /usr/local/rvm/bin/rvm default do gem clean arvados arvados-cli"
-
-ECODE=$?
-
-if [[ "$ECODE" != "0" ]]; then
- title "!!!!!! DEPLOYING ARVADOS GEMS FAILED !!!!!!"
- EXITCODE=$(($EXITCODE + $ECODE))
- exit $EXITCODE
+title "Loading ARVADOS_API_HOST and ARVADOS_API_TOKEN"
+if [[ -f "$HOME/.config/arvados/$IDENTIFIER.arvadosapi.com.conf" ]]; then
+ . $HOME/.config/arvados/$IDENTIFIER.arvadosapi.com.conf
+else
+ title "WARNING: $HOME/.config/arvados/$IDENTIFIER.arvadosapi.com.conf not found."
fi
-
-title "Deploying updated arvados gems complete"
-title "Deploying API server complete"
-
-# Deploy Workbench
-title "Deploying workbench"
-
-# Install updated debian packages
-title "Deploying updated arvados debian packages"
-
-ssh -p2222 root@workbench.$IDENTIFIER.arvadosapi.com -C "apt-get update && apt-get -qqy install python-arvados-fuse python-arvados-python-client arvados-workbench-bare"
-
-ECODE=$?
-SUM_ECODE=$(($SUM_ECODE + $ECODE))
-
-ssh -p2222 root@workbench.$IDENTIFIER.arvadosapi.com -C "/usr/local/bin/arvados-workbench-upgrade.sh"
-
-ECODE=$?
-SUM_ECODE=$(($SUM_ECODE + $ECODE))
-
-if [[ "$SUM_ECODE" != "0" ]]; then
- title "!!!!!! DEPLOYING DEBIAN PACKAGES FAILED !!!!!!"
- EXITCODE=$(($EXITCODE + $SUM_ECODE))
- exit $EXITCODE
-fi
-
-title "Deploying updated arvados debian packages complete"
-
-title "Deploying workbench complete"
-
-# Update compute node(s)
-title "Update compute node(s)"
-
-# Get list of nodes that are up
-COMPRESSED_NODE_LIST=`ssh -p2222 root@$IDENTIFIER -C "sinfo --long -p crypto -r -o "%N" -h"`
-
-if [[ "$COMPRESSED_NODE_LIST" != '' ]]; then
- COMPUTE_NODES=`ssh -p2222 root@$IDENTIFIER -C "scontrol show hostname $COMPRESSED_NODE_LIST"`
-
- SUM_ECODE=0
- for node in $COMPUTE_NODES; do
- echo "Updating $node.$IDENTIFIER"
- run_puppet $node ECODE
- SUM_ECODE=$(($SUM_ECODE + $ECODE))
- done
-
- if [[ "$SUM_ECODE" != "0" ]]; then
- title "!!!!!! Update compute node(s) FAILED !!!!!!"
- EXITCODE=$(($EXITCODE + $SUM_ECODE))
- fi
+if [[ "$ARVADOS_API_HOST" == "" ]] || [[ "$ARVADOS_API_TOKEN" == "" ]]; then
+ title "ERROR: ARVADOS_API_HOST and/or ARVADOS_API_TOKEN environment variables are not set."
+ exit 1
fi
-title "Update compute node(s) complete"
-
-title "Update shell"
+title "Gathering list of shell and Keep nodes"
+SHELL_NODES=`ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN rvm-exec default arv virtual_machine list |jq .items[].hostname -r`
+KEEP_NODES=`ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN rvm-exec default arv keep_service list |jq .items[].service_host -r`
-run_puppet shell ECODE
-
-if [[ "$ECODE" == "2" ]]; then
- # Puppet exits '2' if there are changes. For real!
+nodes=""
+for n in workbench manage switchyard $SHELL_NODES $KEEP_NODES; do
ECODE=0
+ if [[ $n =~ $ARVADOS_API_HOST$ ]]; then
+ # e.g. keep.qr1hi.arvadosapi.com
+ node=$n
+ else
+ # e.g. shell
+ node=$n.$ARVADOS_API_HOST
+ fi
+ # e.g. keep.qr1hi
+ nodes="$nodes ${node%.arvadosapi.com}"
+done
+
+## at this point nodes should be an array containing
+## manage.qr1hi, keep.qr1hi, etc
+## that should be defined in the .ssh/config file
+title "Updating in parallel: $nodes"
+export -f run_puppet
+export -f title
+export SSH_PORT
+export PUPPET_AGENT
+echo $nodes|xargs -d " " -n 1 -P $PUPPET_CONCURRENCY -I {} bash -c "run_puppet {}"
+
+title "Locating Arvados Standard Docker images project"
+
+JSON_FILTER="[[\"name\", \"=\", \"Arvados Standard Docker Images\"], [\"owner_uuid\", \"=\", \"$IDENTIFIER-tpzed-000000000000000\"]]"
+DOCKER_IMAGES_PROJECT=`ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN arv --format=uuid group list --filters="$JSON_FILTER"`
+
+if [[ "$DOCKER_IMAGES_PROJECT" == "" ]]; then
+ title "Warning: Arvados Standard Docker Images project not found. Creating it."
+
+ DOCKER_IMAGES_PROJECT=`ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN arv --format=uuid group create --group "{\"owner_uuid\":\"$IDENTIFIER-tpzed-000000000000000\", \"name\":\"Arvados Standard Docker Images\", \"group_class\":\"project\"}"`
+ ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN arv link create --link "{\"tail_uuid\":\"$IDENTIFIER-j7d0g-fffffffffffffff\", \"head_uuid\":\"$DOCKER_IMAGES_PROJECT\", \"link_class\":\"permission\", \"name\":\"can_read\" }"
+ if [[ "$?" != "0" ]]; then
+ title "ERROR: could not create standard Docker images project Please create it, cf. http://doc.arvados.org/install/create-standard-objects.html"
+ exit 1
+ fi
fi
-if [[ "$ECODE" != "0" ]]; then
- title "!!!!!! Update shell FAILED !!!!!!"
- EXITCODE=$(($EXITCODE + $ECODE))
-fi
-
-title "Update shell complete"
-
-title "Update keep0"
-
-run_puppet keep0 ECODE
+title "Found Arvados Standard Docker Images project with uuid $DOCKER_IMAGES_PROJECT"
+GIT_COMMIT=`ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER "python -c 'import arvados_cwl ; print arvados_cwl.__version__'" 2>&1 |grep -v INFO:rdflib:RDFLib`
-if [[ "$ECODE" == "2" ]]; then
- # Puppet exits '2' if there are changes. For real!
- ECODE=0
+if [[ "$?" != "0" ]] || [[ "$GIT_COMMIT" == "" ]]; then
+ title "ERROR: unable to get arvados/jobs Docker image git revision"
+ exit 1
+else
+ title "Found git commit for arvados/jobs Docker image: $GIT_COMMIT"
fi
-if [[ "$ECODE" != "0" ]]; then
- title "!!!!!! Update keep0 FAILED !!!!!!"
- EXITCODE=$(($EXITCODE + $ECODE))
+run_command shell.$IDENTIFIER ECODE "ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN /usr/local/rvm/bin/rvm-exec default arv keep docker" |grep -q $GIT_COMMIT
+
+if [[ "$?" == "0" ]]; then
+ title "Found latest arvados/jobs Docker image, nothing to upload"
+ # Just in case it isn't yet, tag the image as latest
+ ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER "ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN arv-keepdocker arvados/jobs latest"
+else
+ title "Installing latest arvados/jobs Docker image"
+ ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER "ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN /usr/local/rvm/bin/rvm-exec default arv keep docker --pull --project-uuid=$DOCKER_IMAGES_PROJECT arvados/jobs $GIT_COMMIT"
+ ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER docker tag --force >/dev/null 2>&1
+ # docker 1.13 no longer supports --force. Sigh.
+ if [[ "$?" == "125" ]]; then
+ FORCE_TAG=""
+ else
+ FORCE_TAG="--force"
+ fi
+ ## adding latest tag too refs 9254
+ ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER docker tag $FORCE_TAG arvados/jobs:$GIT_COMMIT arvados/jobs:latest
+ ssh -o "StrictHostKeyChecking no" shell.$IDENTIFIER "ARVADOS_API_HOST=$ARVADOS_API_HOST ARVADOS_API_TOKEN=$ARVADOS_API_TOKEN arv-keepdocker --project-uuid=$DOCKER_IMAGES_PROJECT arvados/jobs latest"
+ if [[ "$?" -ne 0 ]]; then
+ title "'git pull' failed exiting..."
+ exit 1
+ fi
fi
-
-title "Update keep0 complete"
-
-exit $EXITCODE
projects / arvados-dev.git / blobdiff