Hi!

I've been testing this tool but ran into an issue. When I run the tool I can navigate to a task that has ECS exec enabled, but when I select the container to exec into the program panics like this:

? Select a task: panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x1553488]

goroutine 6 [running]:
github.com/tedsmitt/ecsgo/internal.(*ExecCommand).executeInput(0xc00031caa0)
        github.com/tedsmitt/ecsgo/internal/exec.go:264 +0x88
github.com/tedsmitt/ecsgo/internal.(*ExecCommand).Start.func1()
        github.com/tedsmitt/ecsgo/internal/exec.go:72 +0x1e5
created by github.com/tedsmitt/ecsgo/internal.(*ExecCommand).Start
        github.com/tedsmitt/ecsgo/internal/exec.go:58 +0xee

I'm able to work around this by manually specifying the shell to run like ecsgo -c /bin/sh, which according to the README should be the default command.

When trying to tap this. I get the following error

$ brew tap tedsmitt/ecsgo
==> Tapping tedsmitt/ecsgo
Cloning into '/opt/homebrew/Library/Taps/tedsmitt/homebrew-ecsgo'...
remote: Enumerating objects: 21, done.
remote: Counting objects: 100% (21/21), done.
remote: Compressing objects: 100% (14/14), done.
remote: Total 21 (delta 6), reused 3 (delta 0), pack-reused 0
Receiving objects: 100% (21/21), done.
Resolving deltas: 100% (6/6), done.
Error: Invalid formula: /opt/homebrew/Library/Taps/tedsmitt/homebrew-ecsgo/ecsgo.rb
ecsgo: Calling bottle :unneeded is disabled! There is no replacement.
Please report this issue to the tedsmitt/ecsgo tap (not Homebrew/brew or Homebrew/core):
  /opt/homebrew/Library/Taps/tedsmitt/homebrew-ecsgo/ecsgo.rb:10

Error: Cannot tap tedsmitt/ecsgo: invalid syntax in tap!

Add defaults to each select option

Where there is only 1 cluster, task, container then automatically pick that one

This is a starting PR to get comfortable with the code before doing the other SSM changes.

Hello there!

We are having a weird issue using ecsgo. Every time we run the tool and we select a particular container, it logs us into a different one, never to the main container. (In the example below it logs me into the logger container) Per the instructions in your README, I checked the the ssm plugin and the permissions setup for each task. Everything looks good.

I even a run a tool to validate if the ecs exec command and I got the results below.

When i'm in a session and the container get's removed due to new deplyoment/scaling/..., the session hangs forever and the only way is to kill the ecsgo PID.

Is there a possibility to keep track of the websocket state somehow and exit with a message after a certain timeout?

P.S.: Great project anyways, thank you! :heart:

Great utility!

ecsgo doesn't appear to support KMS encrypted sessions, so I changed the command to use aws cli instead.

Before:

✗ ecsgo
? Select a task:
Cluster: redacted | Service: api | Task: 12345308b064a4a911c205c63e0bbbf
Connecting to container redactedi
Starting session with SessionId: ecs-execute-command-12345


SessionId: ecs-execute-command-12345 :
----------ERROR-------
Encountered error while initiating handshake. Fetching data key failed: Unable to retrieve data key, Error when decrypting data key InvalidCiphertextException:

After:

 Select a task:
Cluster: redacted | Service: api | Task: 12345308b064a4a911c205c63e0bbbf
Connecting to container redacted

The Session Manager plugin was installed successfully. Use the AWS CLI to start a session.


Starting session with SessionId: ecs-execute-command-12345
This session is encrypted using AWS KMS.
#

When using strings.Contains our flow breaks when selecting the container - see issue https://github.com/tedsmitt/ecsgo/issues/20 The problem is we have 2 containers that are named like this

• sample-ui-logger
• sample-ui

I pick up the first sample-ui, but it ends up ssh into sample-ui-logger, because .Contains function will match both in that case.

Please check it out, not sure there is a specific reason for using .Contains

As per https://github.com/tedsmitt/ecsgo/pull/11:

ecsgo doesn't appear to support KMS encrypted sessions, so I changed the command to use aws cli instead.

✗ ecsgo
? Select a task:
Cluster: redacted | Service: api | Task: 12345308b064a4a911c205c63e0bbbf
Connecting to container redactedi
Starting session with SessionId: ecs-execute-command-12345


SessionId: ecs-execute-command-12345 :
----------ERROR-------
Encountered error while initiating handshake. Fetching data key failed: Unable to retrieve data key, Error when decrypting data key InvalidCiphertextException:```

What

Add version that will be shown when --version or -v is passed

$ ecsgo --version
ecsgo version Version: v1.0.0, Commit: xxxxxx, Built date: 2021-03-10 18:10:30, Built by: goreleaser

Why

To know which release am I using.

Adds a check to the PlatformFamily value (only available on Fargate tasks) and if it is not available, then we get the container instance OS and add our own "PlatformFamily" value. This prevents the nil pointer issue we were seeing on EC2 launch-type tasks.

Fixes #16

• Add conditional to check the OS family and determine a sensible default command for Windows or Linux
• Rename "cmd" channel to "input" to avoid confusion with the arg "cmd"

Just an idea. Would it be possible to supply cluster, service and taskid via commandline to directly connect to the container without the interactive menu?

e.g.:

ecsgo --profile <profile> --cmd bash --cluster <cluster> --service <service> --task <task>

Merry christmas 😉

It appears that you can port forward to ECS tasks using the AWS-StartPortForwardingSession document via SSM. This purely uses the SSM Client package as far as I can tell, and then you can pass the relevant parameters to the session-manager-plugin.

Will look at getting this implemented asap as I think it'll be very useful.

👋

As mentioned in #21 I noticed that the output when fetching all services on a cluster isn't paginated here: https://github.com/tedsmitt/ecsgo/blob/95178ae50b5d102c5ffea4eb3f5723c6fad40353/internal/exec.go#L124-L125 This causes missing services in the output if you run more than 10 services on a single cluster.

I also looked a bit further in the code and noticed that the same seems to be true when fetching tasks, although if I read the documentation right the ListTasks call seems to return 100 tasks in a single page, so that might not be as much of an issue.

I'd be willing to submit a patch for this, but I can't really say when I'll have the time to do it right now 🙈