Aliyun Ecs Log Analyzer

Aliyun ECS Log Analyzer

Overview

Triage and analyze logs from an Aliyun ECS Linux instance. Collect a time-bounded, read-only slice of system/web/app/auth logs, summarize anomalies (errors, spikes, suspicious IPs), and produce an actionable incident report.

Quick start (ask for these)

• ECS public IP / hostname (and SSH port if not 22)
• SSH user + private key path (preferred), or Cloud Assistant output (fallback)
• Time window (e.g. last 2h / last 24h) and server timezone
• Observed symptoms (e.g. 502 spikes, CPU 100%, login brute-force)
• Stack details: nginx/apache, docker, systemd service name, custom app log path

Default target (local-only)

This skill intentionally does not hardcode a real server. Provide your current ECS target as:

• HostName: <YOUR_ECS_PUBLIC_IP_OR_HOSTNAME>
• Port: 22 (or your SSH port)
• User: <YOUR_SSH_USER> (often root / ubuntu)

Authentication:

• Prefer SSH key-pair login (recommended).
• Do not store passwords, private keys, or any secret tokens inside this skill.

Suggested ~/.ssh/config entry:

Host aliyun-ecs
  HostName <YOUR_ECS_PUBLIC_IP_OR_HOSTNAME>
  Port 22
  User <YOUR_SSH_USER>
  IdentityFile <PATH_TO_PRIVATE_KEY>
  IdentitiesOnly yes

Then test:

chmod 600 <PATH_TO_PRIVATE_KEY>
ssh aliyun-ecs 'echo ok'

Workflow decision tree

1. Prefer SSH access if possible.
2. If SSH is blocked, use Aliyun ECS Cloud Assistant to run the same read-only commands and paste the outputs.
3. If logs are enormous, avoid copying full logs; extract a sample (tail) or aggregate counts (awk/grep).

Workflow: Collect → Analyze → Report

1) Establish context (always)

Run on the server (read-only):

date
timedatectl 2>/dev/null || true
uname -a
cat /etc/os-release 2>/dev/null || true
ps -p 1 -o comm=

Capture current health:

uptime
df -h
free -h 2>/dev/null || true

If you see OOM / reboots, ensure swap is enabled (read the output carefully; this changes system state):

swapon --show
ls -lah /swapfile 2>/dev/null || true
file /swapfile 2>/dev/null || true

If /swapfile exists and file /swapfile shows it is a swapfile, enable it:

swapon /swapfile

Persist across reboot:

grep -qE "^/swapfile\\s" /etc/fstab || echo "/swapfile none swap sw 0 0" >> /etc/fstab

If /swapfile does not exist, create one (example: 4G):

fallocate -l 4G /swapfile
chmod 600 /swapfile
mkswap /swapfile
swapon /swapfile
grep -qE "^/swapfile\\s" /etc/fstab || echo "/swapfile none swap sw 0 0" >> /etc/fstab

If systemd is present, list failed units:

systemctl --failed --no-pager 2>/dev/null || true

2) System log triage

If systemd is present:

journalctl -p warning..emerg --since "24 hours ago" --no-pager | tail -n 200

Otherwise check syslog files (paths vary by distro; see references/log-locations.md):

sudo tail -n 200 /var/log/syslog 2>/dev/null || true
sudo tail -n 200 /var/log/messages 2>/dev/null || true

3) SSH / auth / security triage

Check for brute-force attempts and new logins:

sudo tail -n 200 /var/log/auth.log 2>/dev/null || sudo tail -n 200 /var/log/secure 2>/dev/null || true
sudo grep -E "Failed password|Invalid user|Accepted" /var/log/auth.log /var/log/secure 2>/dev/null | tail -n 200 || true
last -a | head -n 20

If the user is worried about an attack, also list listening ports:

ss -tulpen 2>/dev/null || netstat -tulpen 2>/dev/null || true

4) Web server triage (nginx/apache)

Start with error logs:

sudo tail -n 200 /var/log/nginx/error.log 2>/dev/null || true
sudo tail -n 200 /var/log/apache2/error.log 2>/dev/null || true

For access logs, avoid pasting the full file. Create a small sample on the server, then analyze locally:

sudo tail -n 200000 /var/log/nginx/access.log > /tmp/nginx-access.tail.log

Copy the file (or paste a smaller excerpt) and run:

python3 scripts/nginx_access_summary.py /tmp/nginx-access.tail.log

5) App/service logs (systemd or docker)

For a systemd service:

journalctl -u <service-name> --since "24 hours ago" --no-pager | tail -n 400

For a docker container:

docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}'
docker logs --since 24h <container-name> 2>&1 | tail -n 400

Output format (produce a report)

Write a Markdown report with:

• Time window + timezone
• Symptom summary (what broke, how users are affected)
• Key findings (top errors, top endpoints, top suspicious IPs)
• Likely root causes (ranked)
• Next actions (commands to run, config to check, mitigations)

Safety rules

• Prefer read-only commands and time-bounded queries (--since, tail).
• Ask before running any command that changes state (restart services, rotate/truncate logs, iptables changes, package installs).
• Redact secrets in shared logs (tokens, cookies, passwords, private URLs, internal IPs if needed).

Resources

• scripts/nginx_access_summary.py: Summarize nginx access logs (status codes, IPs, endpoints, request rate).
• references/log-locations.md: Common log locations on Ubuntu/Debian/CentOS and for nginx/docker/systemd.
• references/triage-commands.md: Safe read-only commands for incident triage.

Project ops: Update a systemd service on ECS (sanitized template)

Use this section as a template runbook. Keep it free of real IPs, usernames, repo paths, and key filenames.

On the server:

systemctl status <service-name> --no-pager
systemctl cat <service-name>.service

Common layout (verify on server):

• Working dir: /opt/<service-name>/
• Service: /etc/systemd/system/<service-name>.service
• Env file: /etc/<service-name>.env (contains secrets; do not paste)
• Port bind: 127.0.0.1:<port> (typically behind nginx)

Recommended update method (copy files + restart)

From local machine:

cd <PATH_TO_LOCAL_REPO>

scp -i <PATH_TO_PRIVATE_KEY> -P <SSH_PORT> \
  <FILES_TO_COPY> \
  <SSH_USER>@<YOUR_ECS_PUBLIC_IP_OR_HOSTNAME>:/opt/<service-name>/

Then restart + verify on ECS:

ssh -i <PATH_TO_PRIVATE_KEY> -p <SSH_PORT> <SSH_USER>@<YOUR_ECS_PUBLIC_IP_OR_HOSTNAME> \
  'systemctl restart <service-name> && systemctl status <service-name> --no-pager'

# Recent logs:
ssh -i <PATH_TO_PRIVATE_KEY> -p <SSH_PORT> <SSH_USER>@<YOUR_ECS_PUBLIC_IP_OR_HOSTNAME> \
  'journalctl -u <service-name> -n 80 --no-pager'

Notes / gotchas

• Remote working dir may not be a git repo (files copied via scp).
• Any action that restarts services is state-changing; do it intentionally and verify logs.