When a site is down or slow, check in this order: is it down for everyone or just you, did anything change recently, are you out of a resource (disk, memory, database connections, quota), and what do the error logs say in the minutes before it broke. Most outages trace back to a recent change or an exhausted resource.
1. Confirm what's actually broken
Try from another network or device. Check whether it's the whole site or one page, and whether it's an error, a timeout or just slowness. Note the exact time it started — you'll need it for the logs.
2. Ask what changed
A deploy, a plugin update, a DNS or certificate change, a new integration, a traffic spike from a campaign. If something changed shortly before the problem, rolling it back is often the fastest fix.
3. Look for an exhausted resource
- Disk full — logs and uploads grow quietly until writes fail.
- Memory or CPU — a runaway process or a traffic spike.
- Database connections — every connection used up, so new requests queue and time out.
- Quotas and billing — an expired card or a hit limit on your cloud provider.
4. Read the logs around the start time
Application, web server and database logs from a few minutes before the problem usually show the first error. Fix the first error, not the hundredth.
5. Know when to call for help
If customers can't pay or log in, if data may be at risk, or if you've spent more than an hour without a clear cause, bring in help. Datum Cloud & AI handles rescue, migrations and ongoing DevOps on AWS, Azure and Google Cloud.
Frequently asked questions
How do I prevent the next outage?
Monitoring with alerts, automatic backups you've actually tested restoring, and a simple way to roll back deploys cover most cases.
My site is slow but not down — same checklist?
Mostly. Also check page weight, images and third-party scripts, and whether the database has missing indexes on busy queries.
Should we move to a different cloud provider?
Only with a clear reason, such as cost, a missing capability or reliability. A migration is a project in itself — plan it rather than doing it mid-crisis.