**Security hardening: SSH closed on production servers** (on behalf of @[punk6529])
Done just now: removed "SSH from anywhere on the internet" (port 22, 0.0.0.0/0) from 7 security groups in the main account:
- 6529.io frontend EB (seizeapp-env-node22, both groups)
- GDRC / digital-charter
- 6529-ipfs-server
- seize.io admin portal
- NextGen generator (both groups)
Why: these were open to constant password/key guessing from the whole internet (we saw 100+ attempts a week on one box yesterday).
**How to get a shell now:** every one of these machines already has Session Manager. With your SSO login (LeadDeveloper):
`aws ssm start-session --target <instance-id> --profile <your-sso-profile>`
or EC2 console > select instance > Connect > Session Manager. Port forwarding works too (`--document-name AWS-StartPortForwardingSession`).
**Next (today):** the app ports that are open straight to the internet (3001 on the frontend and GDRC, 4000 on admin, 5001 on NextGen) bypass CloudFront and the WAF. I'll limit them to their load balancers only, one service at a time, checking health after each.
If anything you run (scripts, deploys, CI, a tool on your laptop) SSHes into these, or something breaks, reply here and I'll fix it with you. Rules are backed up and can be restored in a minute.
-
**Update: direct app ports closed, everything healthy** Each change was made one service at a time, checking target health and live URLs before and after: - **6529.io frontend:** port 3001 (and 80/443) no longer open on the instances; only the load balancer can reach 3001. Both targets healthy, EB Green, uncached pages 200 via CloudFront. Direct hits to the instance IPs now time out, so nothing can bypass CloudFront/WAF any more. - **NextGen generator:** 5001/80/443 closed on the instance; only its load balancer reaches 5001. generator.6529.io and generator.seize.io return 200. - **GDRC:** 3001 now only from its load balancer. digitalrightscharter.org and www return 200. - **seize.io admin server:** closed 4000 and 80 (nothing listens on either). 443 left open for now. **Questions for the team:** 1. **admin.6529.io / admin.seize.io returns 502 and has for a while**: the admin app behind nginx isn't running (only nginx on 443; the box has been up 650 days). Does anyone still use the admin portal? If not, we close 443 too and keep the box only as the SSM host for read-only prod DB queries. 2. **6529-ipfs-server** still has 80/443/4001 open. 4001 (IPFS swarm) has to stay public; I'm leaving 80/443 to the IPFS platform work so we don't step on it. If any deploy, script or monitor of yours broke in the last hour, reply here.
-
I haven’t used the admin web portal in ages. Closing 443 is fine with me. Please keep the EC2 instance available through Session Manager for production DB queries.
-
what does it even do anymore? anything relevant? in any case whatever it does we need to migrate
-
it was this repo https://github.com/6529-Collections/6529seize-admin used for a while during drops and allowlists but we moved away from that logic for a while i think both repo and the access to the ec2 can be deleted - i actually shut down the admin service on the ec2 but i still use the ec2 for direct db access to prod when i need to less and less frequently lately but still find myself needing to run some manual db queries sometimes
-
-