Your phone lights up. The website is down, or the shop will not load, or nobody can get their email. When a server stops responding, the pressure to do something, anything, is intense. That is exactly when people restart things at random, change settings they do not understand, and turn a short outage into a long one.
A few minutes of calm, methodical checking usually tells you what has failed and who needs to fix it. Use this checklist in order. You can keep it somewhere handy for the day you need it.
Confirm It Is Really Down
Start by making sure the problem is what it seems to be.
- Try from another connection. Use your phone on mobile data rather than the office network. If it works there, the problem may be your local network or internet connection, not the server.
- Try a different site or service on the same server. If one site is down but others are fine, it is likely a site problem, not a server problem.
- Check your monitoring. If you have external checks, they will show when the problem started and whether it is affecting every location.
Write down the time you noticed it and what you saw, including any error message. “Connection timed out”, “502 Bad Gateway” and “Error establishing a database connection” each point to very different causes.
Narrow Down What Has Failed
Work from the outside in.
Does The Domain Resolve?
dig A example.com +shortIf this returns nothing or the wrong address, the issue is DNS. Check that the domain has not expired and that nobody has changed the nameservers or records.
Does The Server Respond At All?
ping -c 4 203.0.113.10Some servers block ping, so no reply is not conclusive on its own. Try connecting to SSH as well:
ssh [email protected]If neither works, the server may be powered off, frozen or cut off from the network. If SSH works but websites do not, the server is up and a service on it has failed.
Check The Provider’s Status Page
Your hosting or cloud provider will usually post known network or hardware incidents. If there is one, there is little you can fix yourself, and your time is better spent telling customers what is happening.
If You Can Log In
Once you are on the server, check the usual causes. These few commands answer most questions.
Is The Disk Full?
df -h
df -iA full disk, or a full inode table, stops databases, mail and web servers from working. If a filesystem is at its limit, this is your first problem to solve.
Is The Server Out Of Memory Or Overloaded?
free -h
uptime
topLook for very high load compared with the number of CPU cores, swap nearly exhausted, or a single process using most of the CPU or memory. Check whether the out-of-memory killer has stopped something:
journalctl -k --since "1 hour ago" | grep -i "out of memory"Are The Key Services Running?
systemctl status httpd nginx mysqld mariadb php-fpmService names vary between systems, so adjust for yours. Some will not exist on your server, which is fine. A service shown as failed or inactive is a strong lead. Look at why it stopped before restarting it:
journalctl -u mariadb --since "1 hour ago"If You Cannot Log In
Most hosting providers offer a web-based console or out-of-band access that works even when the network or SSH does not. Use it to see what is on the screen. You might find a kernel error, a server stuck at boot asking for a disk check, or a login prompt that shows the server is running but unreachable.
If there is no console and the provider shows no incident, open a support ticket with the details you have gathered: the server, the time it went down, what you tested and what you saw. Clear details get faster answers than “the server is down”.
What Not To Do
- Do not reboot repeatedly. One restart may be reasonable once you have checked the basics, but restarting over and over hides the cause, and a server with a disk problem can fail to come back at all.
- Do not change several things at once. If it starts working, you will not know which change fixed it, or what you broke along the way.
- Do not delete files to free space without knowing what they are. Database files deleted in a hurry cause far worse problems than a full disk.
- Do not ignore recent changes. If something was updated, deployed or reconfigured shortly before the outage, that is the first suspect.
Communicate While You Work
Tell the people affected that you know and are working on it. A short message on social media or a status page reduces support calls and frustration. Keep notes as you go: times, what you checked and what you found. They help whoever takes over if you need help, and they make it much easier to understand the outage afterwards.
Once things are back, find the root cause. A disk that filled will fill again. A service that crashed will crash again. Monitoring and alerts from server monitoring can give you warning next time, and our server and network outages service covers diagnosing and fixing the underlying problem.
How Server Defenders Can Help
If your server is down and these checks have not found the cause, we can help work through it with you and get to the root of the problem. Read about our outage support or contact us through emergency support.
