Allow providing User Data to `ecs up` about amazon-ecs-cli

[ambrons]

I'm looking for this same feature for the purpose of setting up the cloudwatch logging per this documentation:

http://docs.aws.amazon.com/AmazonECS/latest/developerguide/using_cloudwatch_logs.html#cwlogs_user_data

[rms1000watt]

+1 Updates?

[deadanon]

This would save me from manually updating the CloudFormation template of the cluster each time I want to make a minor User-Data change.

[danknittle]

+1 Any update on this?

[PettitWesley]

@2opremio @vultron81 @kaihendry @ambrons @PospolitaNV @komw @codepushr @djones @amit-bridg @deadanon @rms1000watt @danknittle @kristianpaul @schickling @ambrons download 1.9.0 and check out how to use user data with the CLI in our README 😁💃🏾

[vultron81]

+1

[2opremio]

@samuelkarp Would you be open to a PR implementing this?

[samuelkarp]

@2opremio, we're currently looking into how we'd like to support this use-case. If you want to send a PR, we can definitely take a look, but we want to make sure we can support the different types of content that can be set as user data.

[2opremio]

@samuelkarp My plan was just to append whatever was provided to a command line flag (e.g. --extra-user-data) to the existing User Data script.

Since you have bigger plans, I will just stay put on this one.

[kaihendry]

I'm looking for support for this feature since I don't see another way of setting up monitoring agents like http://docs.datadoghq.com/integrations/ecs/ & persistant mounts otherwise.

Else I guess I would have to create forks of amazon-ecs-optimized images, which is a little annoying to do.

[schickling]

+1

[kristianpaul]

+1

[amit-bridg]

how do i add following to ecs.config for clusters created using ecs-cli up

ECS_ENABLE_CONTAINER_METADATA=true
aws/amazon-ecs-agent#709

It seems like simple use cases are not that easily supported and i have to keep hacking :(

[djones]

+1. Specifically need to run this:

echo vm.max_map_count=262144 >> /etc/sysctl.conf

on the host machine to make Elasticsearch happy. https://www.elastic.co/guide/en/elasticsearch/reference/current/vm-max-map-count.html

[codepushr]

+1 Need this for ECS_ENGINE_TASK_CLEANUP_WAIT_DURATION :/

[komw]

+1

[PospolitaNV]

+1

[allisaurus]

Support for providing custom user data was added in 1.9.0. Resolving.

[mailjunze]

NOTE : "--extra-user-data" this flag takes a file name as its argument.
Example :
ecs-cli up --instance-role ecsInstanceRole --keypair jhsgsgs --vpc vpc-jshsgsgb --security-group sg-kjhjkhjgjhgj --subnets subnet-jhgjhg --extra-user-data ud.sh --launch-type EC2 --force

[PettitWesley]

@mailjunze yup that's correct. Do you think the README or command help text should be re-worded in any way? Was it unclear from those that a file argument is needed?

[mailjunze]

@PettitWesley It's mentioned in the README here but not in the official AWS docs: ecs-cli up - Specifying User Data - https://docs.aws.amazon.com/AmazonECS/latest/developerguide/cmd-ecs-cli-up.html#cmd-ecs-cli-up-userdata. I submitted a feedback to update it.

I understand extra-user-data can be shell scripts or cloud-init directives files. Shell scripts are straightforward, However, I am struggling with cloud-init directives files. May I know what should be the content and file extension for cloud-init directives files. An example file would be great.

[PettitWesley]

@mailjunze I'm not an expert on this myself, but I'll try to help. So the file extension doesn't matter, what matters is the first line in your file, this determines what type of file cloud-init will parse/run it as. Valid first line values are things like #!/bin/sh, #include, #cloud-config, #upstart-job, or #cloud-boothook. See this documentation for a full list and explanation. Note that the ECS CLI always packages all of your user data file into a mime-multipart archive; this means that adding one of the valid values as the first line of your file is required. See here for some examples.

Let me know if you run into any issues.