tool.gpt

tools: sys.getenv, get_inst_ip, run_inst_cmd, check_port
You have just received an alarm. The information about the alarm is contained in the following env vars:
ALARM_NAME
INSTANCEID

1. Retrieve the ip of the instance using the id stored in the INSTANCEID env var. Use the getinstip tool providing it the instanceid as an argument.
2. Determine which IP to ssh to by checking connecivity
3. SSH into the instance using the ip and determine the cause of the ALARM_NAME.
4. Try up to 5 diagnostic commands (use sudo)
5. Attempt to resolve the issue
6. Output the diagnosis if any and steps taken to resolve.

---
name: get_inst_ip
tools: sys.exec
description: get an instance ip given an id
args: input: the instance id to lookup

Output both the public and private ip

#!aws ec2 describe-instances --instance-ids "${input}" --query "Reservations[].Instances[].PublicIpAddress" --output json

---
name: run_inst_cmd
tools: sys.exec
description: run a command on a remote instance
args: ip: the instance id to lookup
args: command: the command to run

#!/bin/bash
ssh-keyscan -H ${ip} >> ~/.ssh/known_hosts
ssh -i ~/.ssh/id_rsa ec2-user@${ip} "${command}" || echo "Command failed"

---
name: check_port
tools: sys.exec
args: host: the host to check for connectivity
args: port: the port to check for connectivity

#!/bin/bash

timeout 1 bash -c "</dev/tcp/${host}/${port}" && echo "Port ${port} is open on ${host}" || echo "Port ${port} is closed on ${host}"