问题原文
原帖链接 · Naftuli Kay · CC BY-SA 3.0
At my organization we have a number of simple-to-use base AMIs for different services such as ECS and Docker. Since many of our projects involve CloudFormation, we're using cfn-bootstrap, which consists of a couple of scripts and a service which run on boot to install certain packages and do certain configuration management tasks for us.
On startup of a system, an equivalent of the following script must be executed:
#!/bin/bash
# capture stderr only
output="$(cfn-init -s $STACK_NAME -r $RESOURCE_NAME --region $REGION >/dev/null)"
# if it failed, signal to CloudFormation that it failed and include a reason
returncode=$?
if [[ $returncode == 0]]; then
cfn-signal -e $returncode -r "$output"
exit $returncode
fi
# otherwise, signal success
cfn-signal -s
I was thinking of running this as a systemd oneshot service which runs After=network.target and WantedBy=multi-user.target.
The only problem is that I'd like my AMI to be flexible and only execute this if a certain file exists. Rather than embedding the above script into the EC2 user data, I can have the user data just define an environment file which defines the variables I need and only run my one-shot service if that environment file exists:
#cloud-init
write_files:
- path: /etc/sysconfig/cloudformation
# ...
content: |
CFN_STACK_NAME="stack-name"
CFN_RESOURCE="resource-name"
CFN_REGION="region"
Is there a way to make systemd only run a service if a given condition is met?
已采纳解答
原帖链接 · Michael Hampton · CC BY-SA 4.0
systemd provides a wide variety of conditions you can test. For instance, you can use ConditionPathExists= to test for the existence of a file.
[Unit]
ConditionPathExists=/etc/sysconfig/cloudformation
摘自公开归档,保留原文语言,版式经整理,外部媒体与邮箱地址已移除。
