Server unreachable: what to check before opening a ticket

When a server stops answering, work down this list. Most problems turn up in the first three steps, and each one you do is something we don't have to ask about. If you get to the end, open a ticket and say what you found.

1. Check the server's state#

Open the server in the panel. The header shows its state.

• Running: go on to step 2.
• Stopped: press Start in the header. A server stops when it's shut down from inside (poweroff, shutdown), or by a power schedule.
• Suspended: a renewal wasn't paid. Pay the open invoice under Billing and it starts again as soon as the payment confirms; see Unpaid renewals.
• Rescue: it's booted into rescue mode, which answers on its own SSH login rather than your system. Leave rescue to boot it normally.
• Provisioning or reinstalling, or a key saying what it's doing ("Rebooting…"): wait for it to finish.
• Error: open a ticket and say what you did last.

The Events list on its page also shows anything done to it recently, by you, a teammate, the API or us.

2. Check the status page#

Look at rack4.io/status. If New York or the network shows a problem, we know about it and are working on it, and updates are posted there. You don't need a ticket for a posted incident, but you can open one if your server is still down after it's marked resolved.

3. Open the console#

The Console tab connects to the server's serial console in your browser. It works when the network doesn't, so it tells you which side the problem is on:

• A login prompt (press Enter if the screen stays blank): the server is up. Log in and check its network, firewall and services (step 4).
• Boot errors or an emergency shell: something inside the server stops it booting, often a full disk or a bad /etc/fstab line. Rescue mode lets you mount the disk and fix it.
• Nothing at all, even after a reboot from the panel: open a ticket, and say so.

Logging in on the console needs a password for root or your user; Serial console covers what to do if there isn't one.

4. Check the firewall and services#

Most "unreachable" servers are running, with a firewall or a stopped service in the way.

• The panel's Firewall tab: if the default is deny, a rule must allow the port (22 for SSH, 80 and 443 for the web). See Server firewall.

• The firewall inside the server, from the console:

  bash

  ufw status verbose

  ufw is off on a new server; if you turned it on, make sure it allows SSH. On AlmaLinux and Rocky, check firewall-cmd --list-all instead.

• The service itself:

  bash

  systemctl status ssh nginx
  ss -tlnp

  ss lists what's listening. A service that isn't in the list isn't running, or listens on the wrong address.

• Fail2ban: if you run it, it may have banned your own address after a few wrong passwords. fail2ban-client status sshd shows banned addresses.

5. Check from your side#

• Ping the server's IPv4 address, from your computer and from another network (a phone on mobile data):

  bash

  ping -c 5 203.0.113.42

  It answers on one network but not the other: the problem is between that network and us, not the server. Some office and hotel networks block port 22 too.

• Check the address. A domain may point somewhere old: dig +short example.com should print the server's address.

• SSH errors mean different things: Connection refused means the server answered but nothing listens on the port; Connection timed out means nothing answered. Connect with SSH goes through each one.

Open a ticket with what you found#

If none of that sorted it, open a ticket, pick the server, and set the priority (Urgent if a production server is down). Include:

• what the panel and the console showed (steps 1 and 3); a screenshot of the console is fine, attached to the ticket;
• the output of ping and of the SSH command, with the time you ran them;
• from your side, mtr -rwc 50 followed by the server's address, if the network looks like the problem (see Test your network);
• anything you changed shortly before it stopped answering.

What to include in a support ticket has more on writing one we can act on straight away.