3. Using process substitution

exec 3>&1  # Save stdout
error=$( { aws ec2 delete-snapshot --snapshot-id vid | tee /dev/fd/3; } 2>&1 )
exec 3>&-  # Close fd 3

When working with AWS CLI's JSON output, you might want to parse it properly:

error_json=$(aws ec2 delete-snapshot --snapshot-id vid --output json 2>&1)
if jq -e . >/dev/null 2>&1 <<<"$error_json"; then
    error_message=$(jq -r '.message' <<< "$error_json")
    echo "JSON Error: $error_message"
else
    echo "Non-JSON Error: $error_json"
fi

Here's a complete script example for handling snapshot deletion:

#!/bin/bash

snapshot_id="vid"  # Replace with your snapshot ID or variable
output=$(aws ec2 delete-snapshot --snapshot-id $snapshot_id --output json 2>&1)

if [[ $? -eq 0 ]]; then
    echo "Successfully deleted snapshot $snapshot_id"
    # Process successful output if needed
else
    if [[ $output =~ "InvalidParameterValue" ]]; then
        echo "Error: Invalid snapshot ID format" >&2
        exit 1
    elif [[ $output =~ "InvalidSnapshot.NotFound" ]]; then
        echo "Warning: Snapshot not found - may have been already deleted" >&2
    else
        echo "Unknown error occurred:" >&2
        echo "$output" >&2
        exit 2
    fi
fi

When working with AWS CLI commands in Bash scripts, you might encounter situations where error messages appear in the terminal but aren't captured in variables. This typically happens because AWS CLI writes error messages to stderr (standard error) rather than stdout (standard output).

# This won't capture the error message
result=$(aws ec2 delete-snapshot --snapshot-id vid)
echo $result  # Empty output

In Unix-like systems, there are three standard streams:

• stdin (0): Standard input
• stdout (1): Standard output
• stderr (2): Standard error

By default, command substitution ($(command)) only captures stdout. To capture both streams, you need to redirect stderr to stdout.

Method 1: Redirect stderr to stdout

# Capture both stdout and stderr
result=$(aws ec2 delete-snapshot --snapshot-id vid 2>&1)
echo "$result"

Method 2: Separate Error Handling

# Capture stdout and stderr separately
output=$(aws ec2 delete-snapshot --snapshot-id vid 2> error.log)
error=$(&1 >&3); } 3>&1

Method 3: Using Process Substitution

# More advanced handling
read -r result < <(aws ec2 delete-snapshot --snapshot-id vid 2>&1)

When using --output json, you might want to parse the output properly:

# Example with jq for JSON parsing
result=$(aws ec2 describe-snapshots --snapshot-id snap-123 2>&1)
if jq -e . >/dev/null 2>&1 <<<"$result"; then
    # Valid JSON received
    snapshot_id=$(jq -r '.SnapshotId' <<< "$result")
else
    # Error message received
    echo "Error: $result"
fi

For robust scripting:

#!/bin/bash

set -euo pipefail

delete_snapshot() {
    local snapshot_id=$1
    local output
    
    output=$(aws ec2 delete-snapshot --snapshot-id "$snapshot_id" 2>&1)
    
    if [[ $output == *"InvalidParameterValue"* ]]; then
        echo "Error: Invalid snapshot ID" >&2
        return 1
    fi
    
    echo "$output"
    return 0
}

# Usage example
if ! delete_snapshot "snap-1234567890abcdef0"; then
    exit 1
fi

• Use set -x to debug script execution
• Check AWS CLI version with aws --version
• Verify AWS credentials and permissions
• Test with valid snapshot IDs to see expected output