使用ec2-api-tools获取新启动的实例ID

Question

我通过从简单的bash脚本调用ec2-run-instances启动了一个EC2实例，并希望对该实例执行进一步的操作(例如，关联弹性IP)，为此，我需要实例id。

该命令类似于ec2-run-instances ami-dd8ea5a9 -K pk.pem -C cert.pem --region eu-west-1 -t c1.medium -n 1，其输出如下：

RESERVATION r-b6ea58c1    696664755663    default
INSTANCE    i-945af9e3    ami-dd8ea5b9    pending    0    c1.medium    2010-04-15T10:47:56+0000    eu-west-1a    aki-b02a01c4    ari-39c2e94d   

在本例中，i-945af9e3是我想要的id。

因此，我需要一种简单的方法来从命令返回的内容中解析id -您将如何执行此操作？我的AWK有点生疏了..。您可以在典型的Linux机器上随意使用任何可用的工具。(如果有一种方法可以直接使用EC2-API-tools获得它，那就更好了。但是afaik没有EC2命令来返回最近启动的实例的id。)

Answer 1

好吧，至少像这样的东西应该是有效的：

instance_id=$(ec2-run-instances ami-dd8ea5a9 [...] | awk '/INSTANCE/{print $2}') 

无可否认，我有点懒，认为这样做比重新学习一些AWK基础知识更快……:-)

编辑：按照丹尼斯的建议简化了AWK的用法。此外，为了清晰起见，使用$()而不是，并且去掉了中间变量。

Answer 2

作为正确答案的补充，下面是一个shell脚本，它创建一个实例，运行一些命令并删除该实例。它使用awk的方式与您的相同。

#!/bin/sh

# Creates an Amazon EC2 virtual machine (an instance) and runs some
# shell commands on it before terminating it. Just an example.
# Stephane Bortzmeyer <stephane+amazon@bortzmeyer.org>

# Parameters you can set.
# Choose an AMI you like (ami-02103876 is a Debian "lenny")
AMI=ami-02103876
# Create your key pair first, for instance on the Web interface
KEY=test-b
KEYDIR=.
# The user name to use depends on the AMI. "root" is common but check
# the documentation of the AMI.
USERNAME=root
# Needs to be a legal Unix group of commands
COMMANDS="(uname -a; df -h; cat /etc/debian_version)"
MAX_CONNECTS=4
MAX_TESTS=6

# If you want to change from the default region, set the environment
# variable EC2_URL for instance 'export
# EC2_URL=https://ec2.eu-west-1.amazonaws.com' to use the 'eu-west-1'
# region

# Also, be sure your default security group allows incoming SSH.

if [ "${EC2_PRIVATE_KEY}" = "" ] || [ "${EC2_CERT}" = "" ]; then
    echo "You need to have X.509 certificate and private key locally, and to set the environment variables EC2_PRIVATE_KEY and EC2_CERT to indicate their locations" 1>&2
    exit 1
fi

start=$(ec2-run-instances --key ${KEY} $AMI)
if [ $? != 0 ]; then
    echo "Machine did not start" 1>&2
    exit 1
fi

AMI_E=$(echo "$start" | awk '/^INSTANCE/ {print $3}')
if [ "$AMI_E" != "$AMI" ]; then
    echo "AMI does not match (got $AMI_E instead of $AMI), the machine probably did not start" 1>&2
    exit 1
fi
INSTANCE=$(echo "$start" | awk '/^INSTANCE/ {print $2}')

# I do not find a way to block until the machine is ready. We

# apparently have to poll.
OVER=0
TESTS=0
while [ $OVER != 1 ] && [ $TESTS -lt $MAX_TESTS ]; do
    description=$(ec2-describe-instances ${INSTANCE})
    STATE=$(echo "$description" | awk '/^INSTANCE/ {print $6}')
    NAME=$(echo "$description" | awk '/^INSTANCE/ {print $4}')
    if [ "$NAME" = "" ]; then
        echo "No instance ${INSTANCE} available. Crashed or was terminated." 1>&2
        exit 1
    fi
    if [ $STATE = "running" ]; then
        OVER=1
    else
        # I like bc but 'echo $(( TESTS+=1 ))' should work, too. Or expr.
        TESTS=$(echo $TESTS+1 | bc)
        sleep 2
    fi
done
if [ $TESTS = $MAX_TESTS ]; then
    echo "${INSTANCE} never got to running state" 1>&2
    ec2-terminate-instances ${INSTANCE}
    exit 1
fi
echo "$INSTANCE is running, name is $NAME"
# The SSH server does not seem reachable immediately. We again have to poll
OVER=0
TESTS=0
while [ $OVER != 1 ] && [ $TESTS -lt $MAX_CONNECTS ]; do
    ssh -o "StrictHostKeyChecking no" -i ${KEYDIR}/${KEY}.pem ${USERNAME}@$NAME "${COMMANDS}"
    if [ $? != 255 ]; then
        # It means we connected successfully (even if the remote command failed)
        OVER=1
    else
        TESTS=$(echo $TESTS+1 | bc)
        sleep 3
    fi
done
if [ $TESTS = $MAX_CONNECTS ]; then
    echo "Cannot connect to ${NAME}" 1>&2
fi
ec2-terminate-instances ${INSTANCE}

Answer 3

作为ec2-run-instances的替代方案，您可以创建一个ec2实例，并通过awscli InstanceId逐行获取run-instances

export MyServerID=$(aws ec2 run-instances --image-id AMI --count 1 --instance-type t2.micro --key-name "my_ssh_key" --security-group-ids sg-xxx --subnet-id subnet-yyy --query 'Instances[0].InstanceId' --output text)